Robust State Estimation

Abstract

To determine state information for a subject that is robust to different inputs or conditions.SOLUTION: For drowsiness, facial landmarks can be determined from captured image data and used to determine a set of blink parameters. These parameters can be used, such as with a temporal network, to estimate a state (e.g., drowsiness) of the subject. To improve robustness, an eye state determination network can determine eye state from the image data, without reliance on intermediate landmarks, that can be used, such as with another temporal network, to estimate the state of the subject. A weighted combination of these values can be used to determine an overall state of the subject. To improve accuracy, individual behavior patterns and context information can be utilized to account for variations in the data due to subject variation or current context rather than changes in state.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to robust state estimation. [Background technology]

[0002] There is a constant desire to improve safety in various environments. This can include, for example, accurately determining drowsiness of a person who may be operating a device, such as a vehicle or machinery, that can cause damage or injury if not operated with sufficient care or awareness. While systems exist to attempt to determine states such as drowsiness or alertness, these systems are not accurate in all situations or are not robust enough to variations between users, user states, or environmental conditions. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles” (eg, Standard No. J3016-201806, published on June 15, 2018, Standard No. J3016-201609, published on September 30, 2016, and previous and future versions of this standard) Summary of the Invention [Means for solving the problem]

[0004] Various embodiments according to the present disclosure will now be described with reference to the drawings. [Brief explanation of the drawings]

[0005] [Figure 1A] FIG. 1 illustrates an example of vehicle components, according to at least one embodiment. [Figure 1B] FIG. 1 illustrates an example of vehicle components, according to at least one embodiment. [Figure 2A] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 2B] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 2C] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 2D] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 2E] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 2F] FIG. 1 illustrates feature data that may be utilized, according to at least one embodiment. [Figure 3] FIG. 1 illustrates a drowsiness estimation system that may be utilized, according to at least one embodiment. [Figure 4] FIG. 1 illustrates an exemplary process for estimating a state of a subject, according to at least one embodiment. [Figure 5A] FIG. 1 illustrates a process for accounting for variations or behaviors that are not due to changes in a subject's condition, according to at least one embodiment. [Figure 5B] FIG. 1 illustrates a process for accounting for variations or behaviors that are not due to changes in a subject's condition, according to at least one embodiment. [Figure 6] FIG. 1 illustrates components of a system for determining a person's condition, according to at least one embodiment. [Figure 7A] FIG. 1 illustrates inference and / or training logic, according to at least one embodiment. [Figure 7B]FIG. 1 illustrates inference and / or training logic, according to at least one embodiment. [Figure 8] FIG. 1 illustrates an exemplary data center system, according to at least one embodiment. [Figure 9] FIG. 1 illustrates a computer system according to at least one embodiment. [Figure 10] FIG. 1 illustrates a computer system according to at least one embodiment. [Figure 11] FIG. 1 illustrates at least a portion of a graphics processor according to one or more embodiments. [Figure 12] FIG. 1 illustrates at least a portion of a graphics processor according to one or more embodiments. [Figure 13] FIG. 1 illustrates an exemplary data flow diagram for an advanced computing pipeline, according to at least one embodiment. [Figure 14] FIG. 1 is a system diagram for an exemplary system for training, adapting, instantiating, and deploying machine learning models in an advanced computing pipeline, according to at least one embodiment. [Figure 15A] FIG. 1 is a data flow diagram for a process for training a machine learning model, according to at least one embodiment. [Figure 15B] FIG. 1 illustrates a client-server architecture for extending an annotation tool with pre-trained annotation models, according to at least one embodiment. [Figure 16A] FIG. 1 illustrates components that may be used with a vehicle system, according to at least one embodiment. [Figure 16B] FIG. 1 illustrates components that may be used with a vehicle system, according to at least one embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0006] Techniques according to various embodiments can provide for determining the state of a person or other such subject. In particular, various embodiments provide for determining a person's drowsiness, fatigue, or alertness state based, at least in part, on the person's observed blink behavior over time. An end-to-end drowsiness estimation system can be utilized that includes multiple neural networks and can therefore be relatively robust to changes in input data. One of these neural networks can be used to determine a set of facial landmarks in the captured image data, which can be used to determine a set of eyeblink parameters for the subject over a period of time. This set of eyeblink parameters can be used, such as with a temporal network, to infer a state value for the subject, such as a drowsiness level for the person of interest represented in the captured image data. A separate neural network can determine eye state (e.g., open or closed) from the captured image data without relying on the accuracy of intermediate facial landmarks, which can improve the robustness of the overall process. The eye state information can be used, such as with another temporal network, to infer another state value for the subject. The state values ​​from these temporal networks, each inferred with at least a minimum reliability, may then be used, such as through a weighted combination, to determine an overall state value estimate for the person. This state value may be used to determine whether and what type of action should be taken based on the person's state. To improve the accuracy of such estimates, the system may also attempt to account for behavioral variations between individual subjects, as well as variations due to changes in the current context or environment, such as changes in driving context. Subject and / or context data may be provided as input to the temporal network, for example, so that the network can infer more accurate state data by comparing eyeblink parameters or other observed behavioral data to a baseline that is more relevant for a particular subject under the current context.

[0007] Consider vehicle 100, shown in FIG. 1A , such as a semi-autonomous or computer-assisted vehicle that may include one or more drivers or passengers. In at least one embodiment, vehicle 100 may be a passenger vehicle, such as, but not limited to, a car, truck, bus, and / or another type of vehicle that accommodates one or more passengers. In at least one embodiment, vehicle 100 may be a semi-tractor-trailer truck used to transport cargo. In at least one embodiment, vehicle 100 may be an aircraft, a robotic vehicle, or other type of vehicle.

[0008] Autonomous vehicles may be described in terms of levels of automation as defined by the National Highway Traffic Safety Administration ("NHTSA"), a division of the U.S. Department of Transportation, and the Society of Automotive Engineers ("SAE"), in their respective publications. In one or more embodiments, vehicle 100 may be capable of functionality according to one or more of levels 1 through 5 of autonomous driving. For example, in at least one embodiment, vehicle 100 may be capable of conditional automation (Level 3), highly automated (Level 4), and / or fully automated (Level 5), depending on the embodiment.

[0009] In at least one embodiment, vehicle 100 may include components such as, but not limited to, a chassis, a vehicle body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other vehicle components. In at least one embodiment, vehicle 100 may include a propulsion system 150 such as, but not limited to, an internal combustion engine, a hybrid power plant, a fully electric engine, and / or another propulsion system type. In at least one embodiment, propulsion system 150 may be connected to a drive train of vehicle 100, which may include, but is not limited to, a transmission to enable propulsion of vehicle 100. In at least one embodiment, propulsion system 150 may be controlled in response to receiving a signal from throttle / accelerator(s) 152.

[0010] In at least one embodiment, a steering system 154, which may include, but is not limited to, a steering wheel, is used to steer the vehicle 100 (e.g., along a desired path or route) when the propulsion system 150 is operating (e.g., when the vehicle is moving). In at least one embodiment, the steering system 154 may receive signals from steering actuator(s) 156. The steering wheel may be optional for fully automated (Level 5) functionality. In at least one embodiment, the brake sensor system 146 may be used to operate the vehicle brakes in response to receiving signals from brake actuator(s) 148 and / or brake sensors.

[0011] In at least one embodiment, controller(s) 136, which may include, but are not limited to, one or more system-on-chip (“SoC”) (not shown in FIG. 1A ) and / or graphics processing unit(s) (“GPU”)(s), provide signals (e.g., representing commands) to one or more components and / or systems of vehicle 100. For example, in at least one embodiment, controller(s) 136 may send signals to operate vehicle brakes via brake actuator(s) 148, to operate steering system 154 via steering actuator(s) 156, and / or to operate propulsion system 150 via throttle / accelerator(s) 152. The controller(s) 136 may include one or more on-board (e.g., integrated) computing devices (e.g., supercomputers) that process sensor signals and output operational commands (e.g., signals representing commands) to enable autonomous driving and / or assist a human driver in driving the vehicle 100. In at least one embodiment, the controller(s) 136 may include a first controller 136 for autonomous driving functions, a second controller 136 for functional safety functions, a third controller 136 for artificial intelligence functionality (e.g., computer vision), a fourth controller 136 for infotainment functionality, a fifth controller 136 for redundancy in emergency situations, and / or other controllers. In at least one embodiment, a single controller 136 may handle two or more of the above functionalities, two or more controllers 136 may handle a single functionality, and / or any combination thereof.

[0012] In at least one embodiment, controller(s) 136 provide signals to control one or more components and / or systems of vehicle 100 in response to sensor data (e.g., sensor inputs) received from one or more sensors. In at least one embodiment, the sensor data may be received from, for example, but not limited to, global navigation satellite system ("GNSS") sensor(s) 158 (e.g., global positioning system sensor(s)), RADAR sensor(s) 160, ultrasonic sensor(s) 162, LIDAR sensor(s) 164, inertial measurement unit(s) ("IMU") sensor(s), or other sensors. The vehicle 100 vehicle information may be received from sensors 166 (e.g., accelerometer(s), gyroscope(s), magnetic compass(s), magnetometer(s), etc.), microphone(s) 196, stereo camera(s), wide-angle camera(s) 170 (e.g., fisheye cameras), infrared camera(s) 172, surrounding camera(s) 174 (e.g., 360-degree cameras), long-range camera(s) (not shown in FIG. 1A ), mid-range camera(s) (not shown in FIG. 1A ), speed sensor(s) 144 (e.g., for measuring the speed of the vehicle 100), vibration sensor(s) 142, steering sensor(s) 140, brake sensor(s) (e.g., as part of brake sensor system 146), and / or other sensor types.

[0013] In at least one embodiment, one or more of the controller(s) 136 may receive input (e.g., represented by input data) from the instrument cluster 132 of the vehicle 100 and provide output (e.g., represented by output data, display data, etc.) via a human-machine interface (“HMI”) display 134, an audible annunciator, a loudspeaker, and / or via other components of the vehicle 100. In at least one embodiment, the output may include information such as vehicle speed, speed, time, map data (e.g., a high-definition map (not shown in FIG. 1A )), location data (e.g., the location of vehicle 100 on a map, etc.), direction, the location of other vehicles (e.g., an occupancy grid), information about objects and object status sensed by controller(s) 136, etc. For example, in at least one embodiment, HMI display 134 may display information about the presence of one or more objects (e.g., road signs, warning signs, traffic light changes, etc.) and / or information about a driving maneuver that the vehicle has performed, is performing, or will perform (e.g., currently changing lanes, taking exit 34B in 2 miles, etc.).

[0014] In at least one embodiment, vehicle 100 further includes network interface 124, which may use wireless antenna(s) 126 and / or modem(s) to communicate over one or more networks. For example, in at least one embodiment, network interface 124 may be capable of communicating via Long-Term Evolution (“LTE”), Wideband Code Division Multiple Access (“WCDMA”), Universal Mobile Telecommunications System (“UMTS”), Global System for Mobile communications (“GSM”), IMT-CDMA Multi-Carrier (“CDMA2000”), etc. Additionally, in at least one embodiment, the wireless antenna(s) 126 may enable communication between objects in the environment (e.g., vehicles, mobile devices, etc.) using local area network(s) such as Bluetooth, Bluetooth Low Energy (“LE”), Z-Wave, ZigBee, etc., and / or low power wide-area network(s) (“LPWAN”) such as LoRaWAN, SigFox, etc.

[0015] 1B illustrates an example of camera locations and fields of view for autonomous vehicle 100 of FIG. 1A, according to at least one embodiment. In at least one embodiment, the cameras and their respective fields of view are an illustrative example and are not limiting. For example, in at least one embodiment, additional and / or alternative cameras may be included and / or cameras may be located at different locations on vehicle 100.

[0016] In at least one embodiment, the camera type for the camera may include, but is not limited to, a digital camera that may be adapted for use with components and / or systems of vehicle 100. In at least one embodiment, one or more of the camera(s) may operate at Automotive Safety Integrity Level (“ASIL”) B and / or another ASIL. In at least one embodiment, the camera type may be capable of any image capture rate, such as 60 frames per second (fps), 120 fps, 240 fps, etc., depending on the embodiment. In at least one embodiment, the camera may be capable of using a rolling shutter, a global shutter, another type of shutter, or a combination thereof. In at least one embodiment, the color filter array may include a red, clear, clear, clear ("RCCC") color filter array, a red, clear, clear, blue ("RCCB") color filter array, a red, blue, green, clear ("RBGC") color filter array, a Foveon X3 color filter array, a Bayer sensor ("RGGB") color filter array, a monochrome sensor color filter array, and / or another type of color filter array. In at least one embodiment, a clear pixel camera may be used, such as a camera with RCCC, RCCB, and / or RBGC color filter arrays, to increase light sensitivity.

[0017] In at least one embodiment, one or more of the camera(s) may be used to implement advanced driver assistance system ("ADAS") functions (e.g., as part of a redundant or fail-safe design). For example, in at least one embodiment, a multi-function mono camera may be installed to provide functions including lane departure warning, traffic sign assist, and intelligent headlight control. In at least one embodiment, one or more of the camera(s) (e.g., all of the cameras) may simultaneously record and provide image data (e.g., video).

[0018] In at least one embodiment, one or more of the cameras may be mounted in a mounting assembly, such as a custom-designed (e.g., three-dimensionally (“3D”) printed) assembly, to eliminate stray light and reflections from the interior of the vehicle (e.g., reflections reflected from the dashboard onto the windshield) that may interfere with the camera's image data capture ability. Referring to a door mirror mounting assembly, in at least one embodiment, the door mirror assembly may be custom 3D printed so that the camera mounting plate matches the shape of the door mirror. In at least one embodiment, the camera(s) may be integrated into the door mirror. For side view cameras, the camera(s) may also be integrated into the four pillars at each corner in at least one embodiment.

[0019] In at least one embodiment, a camera with a field of view that includes a portion of the environment ahead of vehicle 100 (e.g., a front-facing camera) may be used for a surround view to help identify the path and obstacles ahead and, with the aid of controller 136 and / or one or more of the control SoCs, to provide information essential for generating an occupancy grid and / or determining a preferred vehicle path. In at least one embodiment, the front-facing camera may be used to perform many of the same ADAS functions as LIDAR, including, but not limited to, emergency braking, pedestrian detection, and collision avoidance. In at least one embodiment, the front-facing camera may also be used for ADAS functions and systems, including, but not limited to, other functions such as Lane Departure Warning ("LDW"), Autonomous Cruise Control ("ACC"), and / or traffic sign recognition.

[0020] In at least one embodiment, various cameras may be used in a front-facing configuration, including, for example, a monocular camera platform including a CMOS (“complementary metal oxide semiconductor”) color imager. In at least one embodiment, a wide-angle camera 170 may be used to perceive objects (e.g., pedestrians, crossing traffic, or bicyclists) coming into view from the periphery. While only one wide-angle camera 170 is shown in FIG. 1B , in other embodiments, there may be any number (including zero) of wide-angle cameras 170 on vehicle 100. In at least one embodiment, any number of long-range camera(s) 198 (e.g., long-view stereo camera pairs) may be used for depth-based object detection, particularly for objects for which a neural network has not yet been trained. In at least one embodiment, the long-range camera(s) 198 may also be used for object detection and classification, as well as basic object tracking.

[0021] In at least one embodiment, any number of stereo cameras 168 may also be included in the front-facing configuration. In at least one embodiment, one or more of the stereo camera(s) 168 may include an integrated control unit with a scalable processing unit, which may provide a programmable logic on a chip ("FPGA") and a multi-core microprocessor with an integrated controller area network ("CAN") or Ethernet interface. In at least one embodiment, such a unit may be used to generate a 3D map of the vehicle's 100 environment, including distance estimates for every point in the image. In at least one embodiment, one or more of the stereo camera(s) 168 may include, but are not limited to, compact stereo vision sensor(s) that may include, but are not limited to, two camera lenses (one on the left and one on the right) and an image processing chip that may measure the distance of a target object from the vehicle 100 and use the generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning features. In at least one embodiment, other types of stereo camera(s) 168 may be used in addition to or instead of those described herein.

[0022] In at least one embodiment, a camera with a field of view that includes portions of the environment to the sides of vehicle 100 (e.g., a side-view camera) may be used for the surroundings view, providing information used to create and update the occupancy grid and generate side collision warnings. For example, in at least one embodiment, surroundings camera(s) 174 (e.g., four surroundings cameras 174 shown in FIG. 1B ) may be positioned on vehicle 100. In at least one embodiment, surroundings camera(s) 174 may include, without limitation, any number and combination of wide-angle camera(s) 170, fisheye camera(s), 360-degree camera(s), etc. For example, in at least one embodiment, four fisheye cameras may be positioned in front, behind, and on the sides of vehicle 100. In at least one embodiment, vehicle 100 may use three surroundings cameras 174 (e.g., left, right, and rear) and utilize one or more other cameras (e.g., a front camera) as a fourth surroundings view camera.

[0023] In at least one embodiment, a camera with a field of view that includes a portion of the environment behind vehicle 100 (e.g., a rear-view camera) may be used for parking assistance, surround view, rear collision warning, and creating and updating an occupancy grid. In at least one embodiment, a wide variety of cameras may be used, including, but not limited to, a camera that is also suitable as a front-facing camera(s) (e.g., long-range camera 198, and / or mid-range camera(s) 176, stereo camera(s) 168, infrared camera(s) 172, etc.), as described herein.

[0024] In various situations, it may be desirable to determine the state of a person in or associated with such a vehicle. This may include, for example, determining the state of drowsiness or alertness of the driver of the vehicle. Approaches according to various embodiments may utilize one or more cameras or imaging sensors, such as those previously described for such vehicles, capable of capturing image or video data of at least a portion of the person of interest. This may include, for example, a driver-facing camera capable of capturing at least the face of the driver or another person seated in the driver's seat or other position associated with at least some amount of vehicle control. Such a camera may be positioned in any suitable location, such as, for example, on or adjacent to the instrument cluster, rearview mirror, left pillar, or touch interface, among other options. Such a camera may be a full-color or grayscale camera of any suitable resolution and, in various embodiments, may also include infrared (IR) or other sensors.

[0025] In at least one embodiment, video or images of such a person may be captured at specific times, periodically, or continuously. If captured at specific times, these times may be determined according to a selection algorithm or may be determined at least in part based on one or more triggers, such as a signal from a vehicle system that is sent in response to an action or behavior that may be associated with driver drowsiness. In other embodiments, a trigger may be sent in response to a change in the environment, such as moving into a crowded urban area where more caution may be required, or the onset of evening or nighttime when the driver may be more likely to become drowsy. Such timing may also be user, application, or device configurable.

[0026] In at least one embodiment, a sequence (e.g., a stream) of images or video frames may be captured using one or more cameras. Under typical circumstances, these images or video frames will include a representation of at least a portion of the face of a person of interest, such as image 200 in FIG. 2A . Depending on factors such as the location and field of view of the camera relative to the person, the captured image may represent primarily the person's facial region or may include a larger view that may include portions of the torso, arms, etc. In at least one embodiment, such an image may be passed to a face detection network or other face detector, which may attempt to determine whether the person's face is present in the image. For example, a face may not be represented if the person is not currently positioned at that location, has turned away from the camera, or is otherwise in a position such that not enough of their face is represented in the image to enable reliable face detection. During vehicle operation, there may be some obstruction blocking the view of the face, such as an arm or hand positioned between the camera and the face.

[0027] If a face is detected in an image (or video frame, etc.), the image may be selected for analysis. For ease of explanation, the terms “image” or “image data” will be used in many instances, but it should be understood that this may also refer to video frames, video data, sensor data, or any other type of data in any format that can capture or represent information for one or more visual aspects of a person. Not every captured image may be analyzed, such as when a person is in a normal or alert state, the system may analyze every 10 images, but when the person is less alert or drowsy, the system may analyze images more frequently. Images may also be analyzed more frequently for certain environmental conditions or driving contexts, as described in more detail elsewhere herein. In at least one embodiment, to reduce the amount of memory and processing capacity required and to focus on more relevant portions of the image, data corresponding to facial regions in the image may be used to select portions of the image to be analyzed, which may help improve the accuracy of inferences or decisions for at least some systems.

[0028] For an image to be analyzed, at least a facial portion of the image may be processed using a first neural network, such as a facial landmark detection network. The facial portion may be identified by coordinates or bounding boxes provided by the facial detection network. Other feature detection mechanisms may be used within various embodiments. The facial landmark detection network may be trained to infer the location of specific types of landmarks, reference points, or features in the image. This may include, for example, identifying any of 100 or more different features represented or detectable in the input image. As shown in FIG. 2A, this may include identifying landmarks 202, such as the top and bottom of a person's ears. Such landmarks may be useful at least because they can help determine the pose or orientation of the head with respect to the camera capturing the image. Other landmarks 232, such as those shown in image region 230 in FIG. 2B, may also be useful. These landmarks correspond to the extremities of each of a person's irises (here, top, bottom, left, and right), which may then be used to determine information such as the amount the person has opened their eyes. Changes in the relative positions of these landmarks 232 between successive frames can be used to detect eye blinks and to determine values ​​such as the amplitude, velocity, or duration of individual blinks. In at least one embodiment, information about any detected landmarks can be output by the facial landmark detection network, including the coordinates of a given landmark in the image, as well as information that may include an identification of the type of landmark identified.

[0029] Data for at least some of these features or landmarks may also be provided as input to other neural networks, algorithms, or modules for determining other aspects or behaviors. For example, facial landmarks related to a person's eyes, along with at least the facial portion of the input image, may be provided to an eye state determination neural network to determine the region of the input image corresponding to the person's eyes. In other embodiments, some of the facial landmarks may be used to crop the input image to only the eye(s), and this cropped image data may be provided as input to the eye state network. This allows the eye state neural network to focus on the portion of the input image corresponding to the eyes to provide more accurate inferences regarding eye states and simplify the corresponding network training process. In at least one embodiment, the eye state network may output a value for each eye state, such as open or closed eye for a binary decision network, or open, closed, or partially open (potentially with a certain value of openness) for others. This may include some amount of openness measurement for the partially open state, such as an inferred distance or percentage of overall eyeblink amplitude, along with a confidence value in this state determination.

[0030] Data for at least some of these landmarks may also be passed to a neural network, process, or algorithm for determining a head pose for the person, at least with respect to the viewpoint of the camera used to capture the image. As shown in FIG. 2A, the relative spacing and location of at least some of these facial landmarks may be used to determine a head pose, such as the roll, pitch, and yaw of the head relative to a default orientation, axis, or coordinate system, such that the head is positioned to appear to be centered in the image and looking directly at the camera, or (if the image data is 2D) orthogonal to the plane of the 2D image. Pose information may be useful for determining differences between landmark positions that are due to head pose or orientation and those that are not due to the person's actions or behavior. Pose information may also be used to normalize the positions of these landmarks or remove the effects of pose. For example, the pose information may be used to determine a transformation to be applied to the determined positions of the landmarks to generate a representation 260 of the landmarks that has the effects of head pose removed, so that the features correspond to the face as if it were in a default orientation or position. Such techniques can help provide more accurate determination of distances between features or landmarks and the variation of those distances over time. In at least one embodiment, facial landmark data can be compared to a three-dimensional (3D) ground truth head model to correlate the landmarks to the model and determine the model's relative orientation in the image. In at least one embodiment, the solvePnP technique can be used to estimate head position and rotation in 3D.

[0031] In some cases, it may be sufficient to utilize a single network rather than both a facial landmark network and an eye state network to determine blink-related information. For example, as shown in image 270 in FIG. 2D , facial landmarks 272 may be utilized to determine when a person's eyes are open, and as shown in image 280, facial landmarks 282 may be used to determine when a person's eyes are closed, as well as intermediate states. These landmarks may then be used to determine information such as blink frequency, in addition to blink parameters such as blink amplitude, duration, and velocity. However, it may occur that a user's face may be in positions, such as "extreme" positions, where these landmarks may not be able to be identified, at least with sufficient reliability or accuracy. As previously mentioned, this may occur when a user is looking away from the camera, the view of the face is partially obscured, the person is wearing reflective glasses, etc. Uncertainty in these landmarks, or intermediate features, may result in uncertainty in blink parameter determination.

[0032] Therefore, an eye state network that can more accurately determine eye states can be utilized, at least under these extreme conditions. The eye state network can be trained in an end-to-end manner so that the eye state network makes eye state inferences directly from input image data without the need for intermediate features or values. As long as a sufficient amount of at least one eye is represented in the input image data, the eye state network can infer eye states with relative accuracy. For blink determination, the open and closed states of the eyes can be determined with relative accuracy so that the overall blink frequency can be accurately determined even if the intermediate values ​​have some uncertainty about the partially open or closed states of the eyes. For example, a plot 290 of eye states (e.g., aspect ratio) by time (e.g., number of frames in a sequence) is shown in FIG. 2F. In this example, it can be seen that many of the measurements correspond to open eye states with values ​​(normalized to 1) close to the maximum open state or blink amplitude. The closed eye state near the bottom of the plot represents approximately the middle of the blink action, or at least the point where the aspect ratio is at its lowest value. Even if the eye measurement for a partially open state is less accurate, the overall shape of the curve will not change substantially, so that the duration between the open states on either side of a blink can be accurately determined. Thus, even in extreme positions or conditions, as long as at least one eye is sufficiently visible in the captured image data, the eye state network can determine the eye state with sufficient accuracy to generate an accurate blink frequency determination. In at least one embodiment, the algorithm for extracting blink features can receive an output from the eye state network indicating when the eye is determined to be open or closed in the current frame. The nearest frame in the sequence can be analyzed to determine whether the consecutive frames detected eye closure.If there are multiple frames with the eyes closed, the smallest eye aspect ratio (EAR) may be determined and designated as the bottom point, as shown in Figure 2F, which may be based at least in part on the facial landmark data. The start and end frames for the blink may then also be determined using the EAR value derived from the facial landmark data.

[0033] As described above, in some embodiments, a binary eye state network can be used in conjunction with a facial landmark network to extract eyeblink features. The binary eye state network can act as a reliable binary classifier and, for a given blink, identify at least one frame in which the eyes are detected as closed. Any of these identified frames, as well as potentially nearby frames in the sequence, can be analyzed to determine or infer a lower point in the aspect ratio determination. The aspect ratio can be calculated based at least in part on the facial landmarks for these frames and thus used to determine a minimum eye aspect ratio for this particular user, which may vary between individuals. The frames in this sequence can then be analyzed in each direction to identify eyeblink state points and blink end points where the eyes are fully open, close to being fully open, or at least at or near a maximum value over a period of time. The sequence can be analyzed to determine a maximum aspect ratio, which can be set as the normal maximum aspect ratio for an open eye state, at least under the current conditions for this particular subject. This value can be used to determine the blink start and stop points, which can help produce more accurate frequency and duration calculations.

[0034] 3 illustrates components of an exemplary drowsiness estimation system 300 that may be utilized, according to various embodiments. In this example, one or more images 302 captured by one or more cameras (or sensors, etc.) facing at least partially toward the person of interest may be provided as input to a drowsiness estimation module 304 (or system, application, service, process, etc.). In the case of a vehicle, this module 304 may be located on the vehicle itself or may be remote, such as on a remote server accessible to the cloud or via at least one wireless network, as described later herein. In other examples, portions of this functionality, such as those related to face detection and / or facial landmark detection, may be implemented on the vehicle to reduce the amount of data that must be transmitted for analysis relative to the full image data.

[0035] In this example, the image data is passed to a face detection network 306. This network may be any suitable neural network, such as a convolutional neural network (CNN), trained to infer the presence of a face in the input image data. If a face is not detected in the image with at least a minimum confidence, the image data may be discarded, and further analysis may not be performed until subsequent data is received. If a face is detected, at least a portion of the image data may be passed to at least a facial landmark determination network 314 for analysis. In at least some embodiments, information regarding the location of the detected face in the image may also be passed along, or the input image data may be cropped to a facial region before providing it as input to the facial landmark detection network 314. As will be apparent, there may be additional components, processes, or modules utilized for at least some of this functionality, such as a module capable of cropping the input image to a facial region given the coordinates of the facial region provided by the face detection network.

[0036] In at least one embodiment, information about the detected face may also be used to attempt to identify a person corresponding to the detected face. This may be performed by the face detection network or by a separate face identification module or network in communication with the face detection network. The input image, detected face data, or other information (potentially including, in some embodiments, facial landmarks determined by the facial landmark detection network 314) may be compared to face data in the person database 308 or other such location. If the person can be identified with sufficient confidence, such as confidence that at least meets a minimum confidence threshold, another determination may be made as to whether there is a profile for that person stored or otherwise accessible in the profile database 310. As described in more detail later herein, such profile information may be used to determine a given user's particular eye blink behavior, which may be used to make more accurate drowsiness estimates by taking into account variability in eye blink behavior between different individuals.

[0037] As shown, image data for at least the facial region, as well as any output from the face detection network 306, and any person profile data, may be passed to one or more neural networks for analysis. In this example, the data is passed to a facial landmark detection network 314, a head pose determination network 312, and an eye state determination network 316, although other networks, algorithms, or processes may be utilized within various embodiments. In this example, the facial landmark network 314 may analyze at least the facial portion of the input image to attempt to infer the location of as many facial landmarks as possible. The network may be trained to recognize any number of different landmarks, such as over 100 different landmarks in at least one embodiment, and may output the location of each detected landmark in the image, the type of landmark, and the confidence in its determination or inference.

[0038] In at least one embodiment, at least a portion of the facial landmark data may be provided to the head pose determination network 312, potentially along with the input image data. The head pose determination network may be trained to infer head pose or orientation data, such as roll, pitch, and yaw, based at least in part on the relative positions of facial landmarks in the input image. In at least one embodiment, the head pose network may output inferred values ​​for each of the roll, pitch, and yaw, or other such orientation determinations, along with one or more corresponding confidence values. In some embodiments, the head pose determination network may be an end-to-end network that infers head pose from the input image data without using facial landmark data, which would make the head pose determination network more robust; however, since facial landmarks are determined in this system anyway, these landmarks may be used to more efficiently determine head pose with similar accuracy under most conditions. If head conditions prevent the landmarks from being accurate, the head pose information may not be used anyway, as described elsewhere herein.

[0039] The head pose information, along with the facial landmark data, may be provided to the blink parameter determination module 318. In at least one embodiment, this module may use the head pose information to normalize the facial landmark data or remove variations in relative distances between landmarks due to facial orientation in the input image data. The blink parameter determination module may then use data from this image, as well as previous images in the sequence, to determine values ​​for various blink parameters. This may include determining blink amplitude, velocity, and duration over a recent time period, such as the last 10 seconds, 30 seconds, or minutes. Values ​​for other blink or state parameters may also be determined within various embodiments. Furthermore, while these blink parameters may be handcrafted and easily understood, at least some of these parameters may be previously undetermined parameters that are learned by the network during training. In this example, these eyeblink parameter values ​​may then be fed into a long short-term memory (LSTM) network 322, a transformer, or a temporal network such as a gate recurrent unit (GRU), or a support vector machine (SVM) for analysis. The eyeblink parameter values ​​may be concatenated with other drowsiness indicator signals, such as steering patterns (e.g., amplitude, frequency, or standard deviation of steering movements, turns, etc.), lane keeping patterns (e.g., number of lane crossings, standard deviation of lane position), and electroencephalography / electrocardiography (EEG / ECG) patterns, as input to the LSTM network 322. The LSTM network can apply different weights to eyeblink values ​​determined at different times, such as by applying a greater weight to events in the recent past than in the distant past, and can use this information to infer a drowsiness value for the person.Weighting recent data more heavily helps the system be more robust to changes in environmental data or driving context over time, or changes in the user's physiology, while still taking into account observed patterns over time for a given person. As described later in this specification, this may include determining a first state value, such as one that may correspond to a Karolinska Sleepiness Scale (KSS) value for drowsiness state determination. Other state values ​​may also be determined for drowsiness-related state determination, such as any drowsiness scale value (e.g., a value on the Stanford Sleepiness Scale or the Epworth Sleepiness Scale), or a value on any scale that represents fatigue due to drowsiness. Such values ​​may also be generated for any measure (e.g., subjective or well-functioning objective) that represents a loss in performance or ability, such as a loss of driving performance due to drowsiness or fatigue (or other condition of interest). In at least some embodiments, a temporal network such as an LSTM may provide more accurate results than a model or network such as an SVM, which may be used to attempt to determine drowsiness instantaneously, instead of analyzing one or more patterns over a period of time based on multiple eyeblinks.

[0040] As mentioned above, the drowsiness value may not always be reliable, especially for extreme conditions such as a head facing away from the camera during image capture. To provide robustness of the drowsiness estimation for such situations, a second estimate of drowsiness may be performed using the eye state network 316. In this example, the eye state network 316 may receive facial landmark data from the facial landmark network for at least landmarks (or bounding boxes, etc.) related to the person's eyes. The eye state network may then use this information to focus only on one or more portions of the input image that represent the person's eye or eyes. In other embodiments, the eye state network may analyze the input image, or at least the facial region of the image, without receiving facial landmark data. The eye state network 316 may be trained to determine the state of the eyes in an image, such as whether the eyes are fully open, fully closed, or partially open / closed. In at least some embodiments, the network may infer a value representing "open" or "closed," such as a value between 0 for fully closed and 1 for fully open. The "sufficiently open" eye may correspond to the eye that is open to the maximum observed eyelid separation or otherwise possible for this user. The network may output such values ​​for each detected eye or for both eyes together, along with a respective confidence value for each determination. In at least one embodiment, this information may be provided as another input to the blink parameter determination module 318 for use in making more accurate blink parameter determinations. In this example, the eye state information, along with relationship information over a recent time period, may be provided to the blink frequency determination module 320, system, device, process, or service. The blink frequency determination module 320 may analyze changes in eye state over a recent time period to determine the occurrence of a blink action by the person, which may be used to calculate a blink frequency value over the recent time period.This blink frequency information, along with blink frequency information for previous time periods in this example, may be provided to another LSTM network 324, which applies different weights to blink frequency data obtained for different periods over the recent past. The frequency information may be concatenated with other drowsiness indicator signals, such as steering patterns (e.g., amplitude, frequency, or standard deviation of steering movements, turns, etc.), lane keeping patterns (e.g., number of lane crossings or standard deviation of lane position), EEG / ECG data, etc., as input to the (LSTM) network 324. This LSTM network 324 can then infer a second drowsiness value for the person, which may correspond to determining a second KSS value. While in some embodiments, these values ​​may be provided separately, in this example, these values ​​may be used to provide a single-output drowsiness estimate. Furthermore, if the confidence value for the facial landmark data does not meet at least a minimum confidence threshold, such as in the case of extreme head positions or images with partially obscured faces, the facial landmark data may not be provided to the first LSTM network, and only the second LSTM network may be used to make a drowsiness prediction for that image based on one or more results of the eye state network.

[0041] If the somnolence values ​​agree, at least within a tolerance of agreement, that estimate may be provided as the output somnolence value 330. If the values ​​do not agree, the relative reliability values ​​for the separate decisions may be analyzed. If one decision has high reliability but the other does not, the decision with high reliability may be provided. In other examples, the value may be a blend of the two decisions weighted by reliability; thus, if one score of 8 has high reliability and one score of 5 has low reliability, a final result of 7 may be determined based on the higher weighting applied to the score of 8. Various other approaches may also be used within the scope of various embodiments.

[0042] As described above, values ​​can be inferred or generated that can provide a view of a person's state at a particular point or period in time. While values ​​such as KSS values ​​have previously been generated to attempt to provide a measure or indication of a particular state, conventional approaches often did not provide sufficient accuracy and robustness due to factors such as inaccurate feature detection. Systems according to various embodiments can overcome such deficiencies by utilizing multiple networks, where the networks can utilize or rely on different information and are therefore not limited by a single set of inaccurate features or values. As described above, this can include the use of an eye-opening / closing network for eye-blink detection, which can be used to derive eye-blink frequency features (e.g., PerCLOS), and a facial landmark network for blink feature extraction, including eye-blink amplitude, blink duration, and blink velocity. Such a system can also include two temporal (e.g., LSTM) networks to correlate eye-blink events and features with the KSS. Based at least in part on facial landmark reliability and head pose data, the drowsiness estimation system can select high-confidence features to train these temporal networks. If the facial landmark reliability is high, the temporal network can be trained using blink frequency features along with blink amplitude, blink duration, and blink rate, and these features can be frontalized (or the influence of orientation on position can be removed) using head pose data. Alternatively, if the facial landmark reliability is low, the temporal network can be trained using only blink frequency data. Such an approach can significantly improve drowsiness estimation accuracy through the use of multiple networks and robust feature selection.

[0043] 4A illustrates an exemplary process 400 for determining drowsiness that may be utilized, according to various embodiments. For this and other processes described herein, unless otherwise specified, it should be understood that there may be additional, fewer, or alternative steps performed in a similar or alternative order, or at least partially in parallel, within the scope of various embodiments. Furthermore, while this example is described with respect to drowsiness of a person operating a vehicle, the determination of a person's state may also be used for other types of states of people or subjects performing other types of activities, within the scope of various embodiments. In this example, image data including a representation of at least a portion of the face of a person of interest (or other subject), such as a vehicle driver, is received 402. This image data may be analyzed using a face detector network to determine information regarding the presence of a face as well as the location of the face in the image 404. If a face is not detected, or if a face cannot be determined with at least a minimum confidence, the process may discard this image data and await subsequent image data in which a face may be detected. If a face is detected, at least the facial region of this image data may be analyzed 406 to attempt to identify a number of facial landmarks that may be useful in determining state information about the person, such as drowsiness.

[0044] Once at least a determinable subset of facial landmarks has been identified, these facial landmarks may be used with several different tasks. One such task involves providing at least some of these facial landmarks to a head pose network to attempt to determine the head pose of the person of interest, as represented in the image data 408. In other embodiments, the head pose may be determined directly from the input image data. In this example, the facial landmark data is also used to crop the input image data or identify at least a portion of the input image data to primarily include one or more eye regions, such as one region for both eyes (if represented in the image data) or separate regions for each eye 410. The image data for these one or more eye regions may be provided as input to an eye state determination network, which can determine the state of the person's eyes, such as open or closed for a binary network, or partially open for a non-binary network. Additionally, any or all of the head pose, eye state, and set of facial landmarks may be utilized to determine values ​​for a set of eye parameters, such as blink rate, duration, and amplitude 412. These eye parameter values, along with similar values ​​over a recent time period, may be provided as input to a first temporal network to attempt to infer a state value, such as a drowsiness value, for the person of interest 414. In some embodiments, this step may be performed only if a set of facial landmarks and / or state parameters can be determined with at least a minimum level or threshold of confidence. In parallel, eye state data from the eye state determination network may be provided to another temporal network 416, along with similar data over a recent time period, to infer another drowsiness value based on eye state data that is independent of intermediate facial landmark data, which may make this determination more robust to variations in input image data. An overall drowsiness value for this person may then be determined 418 based at least in part on these inferred drowsiness values, if available.This may include, for example, performing a weighted average or selecting the most reliable somnolence value, among other options. Once the overall somnolence value is determined, the system, service, application, module, or process receiving this value may determine whether to take action based at least in part on this overall somnolence (or other such condition) value 420. This may include taking corrective or proactive action, such as, for example, providing a notification, generating an alarm, or at least in part controlling a current task being or to be performed by the person.

[0045] Such techniques can be highly accurate in estimating a person's current state. However, different people often exhibit different behaviors, such as different blink behaviors. To improve the accuracy of state estimates for different subjects, it may be beneficial to attempt normalization across users in the final few examples. Different individuals may have different base levels or ranges of how certain characteristics or behaviors are manifested, which may be linked to physiological states in those individuals. Different individuals may also have different patterns of eye movement, such as different blink patterns and rates. Somnolence and fatigue determinations are quite complex and may be based on several different factors. Because a person's blink rate may increase at the onset of somnolence, with a corresponding decrease in blink rate or increase in blink duration, a good approximation of somnolence and fatigue can be obtained by analyzing a set of blink parameters. The person's blink amplitude may also decrease at the onset of somnolence. Any of these blink parameters, individually or in combination, may be used to estimate a person's or subject's somnolence. However, for this estimate to be accurate, it may be beneficial to determine how the values ​​of these parameters change for individual subjects in different states of drowsiness. Because some people blink much more frequently than others, and potentially at different blink rates, regardless of whether they are drowsy or not, there may be a significant amount of variability between subjects. Therefore, a global threshold or assessment may not provide accurate results for all observed users, as a blink frequency that is normal for one person may represent a state of drowsiness for another. Therefore, various exemplary approaches may attempt to learn or acquire information about an individual's actions or behavior in order to make a more accurate estimate or assessment for that particular individual.

[0046] As described above, various techniques can be used to identify the person for whom state estimation should be performed. In some embodiments, this can include a user logging in or selecting a profile from an interface in the vehicle. For some vehicles, there can be other ways to identify the person, such as through various biometrics, or the person can be set as the default operator of the vehicle. In some cases, identity can be determined through the use of a specific key or input. Also, a facial (or other body part) identification or recognition system can be used as described above, such that at least one image of a portion of the person is captured and then analyzed to attempt to determine identity. This can include, for example, comparing facial features to a feature set stored in a user database, among other options. Once a person is identified, information about that person can be used to improve state estimation. This can include, for example, retrieving behavioral data from a user profile or determining a behavioral category or type for the person, among other options.

[0047] A user profile (or other data repository for one or more people) can contain various types of data that may be relevant for one or more types of state estimates. This can include, for example, data such as average or “normal” eyeblink rate or blink amplitude, as well as the general range of such values. This allows estimates about states such as drowsiness to be made relative to a baseline that is accurate for that particular individual. Data about other actions, characteristics, or patterns may also be stored, such as those that may relate to different driving skills or behaviors, different reaction speeds, etc. for a particular individual, such as those that may relate to how often a person adjusts the steering wheel or the pattern a person follows for changing lanes. In at least some instances, these values ​​may be vehicle-dependent; a person may have different values ​​for different vehicles, such as for a sports car versus an SUV, which may inherently have quite different handling or operating characteristics. In at least one embodiment, the robustness of a state estimation system or service can be improved by accounting for population variation, such as by taking individual details into account during model training.

[0048] In at least one embodiment, a somnolence model for a person can be trained using data about that individual. This can include, for example, utilizing self-reported KSS or state data as well as data from a subject-specific profile. Data from the subject profile can include information such as driving experience level, age, gender, nationality, or any individual differences that may cause or be correlated with state variations, such as somnolence symptoms. The subject profile can also include baseline signal values ​​corresponding to when the subject is in a particular state, such as a particular level of somnolence, which may relate to a particular blink rate or range of blink rates, or behavior when in an alert state versus a drowsy state. The profile baseline signal can be used to normalize the data so that results produced using the somnolence estimation model are not significantly affected by individual differences. For example, two subjects may have different blink rates while in an alert state and therefore may have different blink rates when in a drowsy or sleepy state. Without a profile baseline or other such subject-specific data, a model may have difficulty learning thresholds or ranges for distinguishing states, such as whether a subject is sleepy based on absolute eyeblink rate values. A profile baseline need not be derived from a specific state, such as an alert state, but may also or alternatively be derived from any level of somnolence. In at least one embodiment, multiple somnolence estimation models may be trained based on different profile baseline data for any or all somnolence levels, such that a model can estimate somnolence given a baseline profile at any somnolence level. In at least one embodiment, a feature vector containing eyeblink parameters generated by or determined from the output of a facial landmark network may be normalized before being passed to an LSTM network for analysis. This may include, for example, taking the feature vector and using it as the mean, or normalizing it with respect to a range of features, subtracting the mean, and dividing by the standard deviation of the feature vector.

[0049] In one embodiment, when a subject enters a vehicle or is present in a monitored location, such as the driver's seat, the system may attempt to identify the person. As described above, this may include using facial recognition or biometrics to identify the person. If identified, the system may then attempt to determine whether the person or subject has an available profile, whether stored on a non-transitory storage medium in the vehicle or accessible via at least one network. If such a profile is available and accessible, the system may utilize the data in the profile to normalize data about the user and implement an appropriate state estimation model. If the subject does not have an existing and accessible profile, an attempt may be made to generate such a profile. In some embodiments, this may include monitoring or capturing information related to the user over a period of time to attempt to determine baseline information. In some embodiments, the person may have the option to indicate whether they want a user calibration process to be performed, which may allow the person to provide certain information and have other information captured or obtained that may be useful in generating one or more state baselines for user activity or behavior.

[0050] To provide accurate estimates, the camera or sensor may also be operated at a relatively high capture rate. A user's blink duration may be a fraction of a second, and therefore, it may be beneficial to operate the camera at a rate of at least 30 Hz or 60 Hz to obtain sufficient data about the blink. For example, if a person's blink duration is approximately 0.2 seconds, it may be necessary to operate at a capture rate of at least 30 Hz to obtain at least 5-6 data points about the blink; fewer points may result in greater uncertainty in the timing of the blink's onset, middle, and end for purposes of determining values ​​such as blink rate and duration. In at least some embodiments, it may be desirable to make this data collection as unobtrusive as possible, so that the user may not even be aware that data is being collected, and for activities such as driving, data collection would in no way reduce the safety of the activity.

[0051] If a person or subject elects to perform user calibration, the person may be instructed to follow a specific data collection procedure to provide one or more rounds of data for the system to use in estimating a profile baseline specific to that subject. The subject may not need to provide a complete set of data in this data collection step, as a user may not be able to provide data for every possible user state. The system can take the data provided for several states and use this data to locate the closest profile in a profile database, for example, using a nearest neighbor search process, locate data to use to complete the subject's profile, or at least use it to infer any missing data. As part of the calibration process, the system may allow for user feedback or input. If the subject finds that the somnolence estimate deviates significantly from their self-reported KSS or current perception of somnolence, the subject can provide feedback that can be used to adjust and / or retrain the associated model or adjust one or more baseline values ​​or ranges for the subject.

[0052] If the person chooses not to perform user calibration, the system can attempt to identify a profile that may approximate this particular person's behavior. This may involve analyzing the captured image data to attempt to determine certain aspects of the person, such as age, gender, breathing pattern, or heart rate, and can select the profile that most closely matches people with at least some of those aspects.

[0053] FIG. 5A shows an example process 500 for accounting for user variation in behavioral data. In this example, image data including a representation of at least a portion of a face of a person of interest is received 502 (or otherwise obtained). In other examples, other information about the person that may help identify the person, such as may relate to biometric, identity, or other such information, may be received. In this example, the image data is analyzed 504 using a facial recognition process to determine the identity of the person of interest represented in the captured image data. A determination may then be made 506 as to whether an associated behavioral profile exists and is accessible for the person. If no identification could be made, such a profile would also not be able to be determined. If such a profile is determined to exist 508, associated behavioral data for the person may be obtained from the profile 510. If such a profile does not exist or is otherwise not accessible, the user may be requested or prompted to undergo or participate in a calibration process to attempt to gather information useful in creating such a profile for the person. If it is determined at 512 that such a calibration process can proceed, the calibration process may be performed to generate a behavioral profile for the person 514. As described above, this may include collecting image and sensor data for the person and, potentially, receiving user input regarding conditions to generate calibration, pattern, or baseline data for the person under at least some conditions or contexts. Similar profiles may be examined to attempt to fill any gaps in the profile or provide a starting point for calibration. If calibration is not to be performed or the person is not identified, a closest or default profile may be selected based on any available and allowed data that may relate to the person, location, or aspects of the action to be performed.In some examples, if a user cannot be identified or refuses to perform calibration, such a process may decide to avoid using user behavioral data to attempt to improve state determination accuracy. If a profile is determined or generated, behavioral data from the profile may be utilized to normalize the state data to account for variations between individual users 518. This may include providing the behavioral data as input feature vectors to a temporal network that will infer state values ​​based on observed inputs about the user, such as eyeblink parameter values ​​determined over a recent time period, and person-specific baselines, ranges, or behaviors may be used to more accurately infer a state for this person.

[0054] As mentioned above, a person's behavior or actions may also vary depending on other factors that may relate to the environment or context in which the person is located or involved. In the case of a person operating a vehicle, this may include various types of driving context information, which may include various environmental factors and other such factors. For example, a person may blink or squint more frequently when in a bright environment than when in a dark environment. Furthermore, a driver may tend to pay more attention and move their eyes more frequently in a crowded urban location than in a rural setting with few other vehicles or objects nearby. At least some of these environmental conditions may be determined by analyzing image or video data captured by one or more cameras associated with the vehicle, such as those shown in FIG. 1B. Environmental data may also come from other sources, such as external data sources (e.g., traffic, weather, or navigation services) accessible via at least one network, an internal clock, or a temperature sensor. Other data acquired by the vehicle's sensors, such as some of those described above with respect to FIG. 1A, may also be used to determine various environmental conditions or aspects of the current driving context, such as whether the road is curved or straight based on steering wheel movement, whether there is heavy traffic or a stop based on braking information, etc. If GPS or navigation data is available, that information may also be used to provide at least some amount of contextual information about the current route or location. Brightness sensors may be used to determine lighting conditions, vehicle control systems such as steering and braking control systems may be used to infer aspects of road or traffic conditions, etc.

[0055] Such driving contexts can be used to improve the baselines used for state inferences or estimates about a person operating within that driving context. Any or all available context information (or other relevant information or inferences) can be analyzed to attempt to determine one or more state baselines for the person under current conditions. This can include, for example, adjusting a range or baseline that can be used to infer a different state of drowsiness based on these conditions. For example, if a person is likely to blink more frequently during a snowstorm than under normal conditions, adjusting the baseline for this driving context can help prevent this increase in blink frequency from being interpreted as a change in drowsiness. Similarly, subject behavior that may be considered normal and typical may actually indicate that the user is becoming drowsy if the user behaves differently under the current driving context, and failing to consider the driving context can cause this increase in drowsiness to potentially go undetected.

[0056] In some systems, driving contexts may be applied similarly across all users. In other systems, individual profiles may be updated with information about different driving contexts. This may include, for example, monitoring user behavior for different driving contexts over time and updating the baseline based on the observed behavior. In some systems, the user may allow the collection of information that can help calibrate the system, such as by responding to questions about drowsiness or other conditions while the vehicle is operating in a particular set of driving conditions or contexts. As with individual profiles, there may not be enough information to fill in complete profiles for various driving contexts; therefore, the system may attempt to fill in missing information by drawing information from one or more profiles for other subjects with similar driving context variations.

[0057] The use of driving context (or similar types of environmental data for other activities) can enable a state estimation system to generalize well when tested on different individuals under a variety of different driving contexts. To further improve accuracy, such systems can utilize inputs from multiple sources to obtain a more accurate view of user behavior as well as the current driving context or environmental conditions. Many conventional systems that attempt to determine drowsiness rely on data from only one source, such as a single sensor or method, and do not consider any variations in driving context. These systems may look for user actions, such as yawning in video data or changes in steering patterns, as the only indicator. Because drivers with the same drowsiness level but in different driving contexts have different physiological behaviors and driving performances, detection systems that use only one of these methods without driving context risk low accuracy because these drowsiness signals depend on the driving context. By fusing information from multiple sources and considering driving context in addition to user variations and behavior, a robust drowsiness estimation system that is accurate across different subjects and conditions can be provided.

[0058] As described above, for a vehicle, various cameras, sensors, or data sources may be used to obtain information about the current driving context. Similar sources may be used for other types of activities. For driving context, this data may include data related to road curvature, traffic volume or type, lane marker visibility, time of day, season, weather, lighting conditions, speed, configuration, type of driving lane, road surface type, wind speed, window conditions, radio conditions or volume, interior lighting, or other data related to environmental aspects that may trigger detectable changes in subject behavior, such as eyeblink behavior. As described above, if a profile is not available for a subject or the subject cannot be identified, a general driving context profile may be used or selected based at least in part on determinable (and permissible) aspects of the subject, such as age, gender, health, or region, that may affect driving behavior. At least in part, because such data may be sensitive to collecting or using to make a determination (or, in some cases, not permissible to collect or use to make a determination), other methods may be used to attempt to predict driving behavior for a user, such as when information about the driver's experience level or familiarity with a particular vehicle can be obtained. Any of this information may be useful in improving profile baselines for normalizing data, allowing somnolence or state estimation models to be minimally affected by variability between users or situations. In at least some embodiments, multiple somnolence estimation models may be trained based on different profile baselines at all somnolence levels, such that a model can estimate somnolence given a baseline profile at any somnolence level. Robustness is obtained, at least in part, because driving context affects physiological behavior and driving performance, and these effects may further vary between subjects.For example, congested road driving conditions may require more driver attention than driving on an empty road, which may result in increased cognitive load, decreased blink rate, and increased gaze / saccade fixation, among other potential changes. Also, aspects such as road curvature, weather, route familiarity, and lighting conditions, as well as driving patterns such as how a driver adjusts the steering wheel and stays within a lane, may affect physiological behavior. For example, even for the same route under similar conditions, a user may behave differently if they are driving the route for the first time than if they have taken the route frequently and are familiar with the route, and therefore may tend to pay less attention or be less vigilant.

[0059] In at least one embodiment, the context monitor 334 may be used to analyze available or received data 332 relating to a driving context or other set of environmental conditions, as shown in FIG. 3 . This may include, for example, analyzing input data to determine or infer a current driving context from among a set of possible driving contexts. In other embodiments, this may include determining a set of context inputs or features that may be used to select an appropriate baseline from an appropriate model. This context monitor may operate continuously, periodically, or upon detection of a significant change in at least one context input, among other options. In some embodiments, the context monitor may include a neural network trained to determine a driving context given a set of inputs. In other embodiments, one or more algorithms or processes may be used to determine one or more context factors or values ​​from the obtained inputs. This driving context may then be used to determine, set, select, or calculate an appropriate baseline or range for one or more behaviors or conditions to be monitored under current conditions.

[0060] As with the user profile described above, when a subject enters a vehicle or is present in a monitored location, such as the driver's seat, the system can attempt to identify the person. If identified, the system can then attempt to determine whether the person or subject has an available profile. If such a profile is available and accessible, the system can utilize the data in that profile to normalize data about the user for the current driving context and implement an appropriate state estimation model. If the subject does not have an existing and accessible profile, an attempt can be made to generate or acquire such a profile. In some embodiments, this may include monitoring or capturing information related to the user over a period of time to attempt to determine baseline information and deriving missing information from similar profiles.

[0061] In some embodiments, blink parameters may be provided in time increments, such as for the most recent minute in time. This may include, for example, the number of blinks determined over that minute, as well as information about those blinks individually or collectively. A history of the determined quantities may be kept in a history buffer, such as for the last 60 minutes or last 60 periods of blink data, in which case the input to the LSTM network may be a feature vector 60 features in length. Any driving context or user profile information may also be provided as a vector of a similar length, or otherwise as an input to at least one of the LSTM networks. Maintaining a history of driving context or user physiological state over time can help make more accurate decisions based on blink or condition behavior over that same time period. In at least some embodiments, there may be different vectors provided for different types of context information. In at least some embodiments, the feature vector (or configuration of a given feature vector) may vary over time based at least in part on the type or amount of context or condition data available at that time. These feature vectors for the user profile and driving context data can then be used to adjust thresholds or otherwise normalize the eyeblink parameter data to account for expected variations (not related to somnolence) and attempt to detect variations in eyeblink behavior that are due solely to changes in somnolence.

[0062] FIG. 5B shows an example process 550 for improving the accuracy of state estimation by considering the current action context or other environmental or condition information. In this example, action-related context data may be received from multiple sources 552. In the case of a driving action, this may include any driving-context-related data, such as image data captured from a camera on the vehicle, sensor data about the vehicle, operational information about the vehicle, environmental data from a network-connected service, or location data, among other options, as described and suggested elsewhere herein. In some embodiments, a superset of information may be received or obtained, and a relevant subset of that data may be determined or selected. This received data may be used to generate 554 one or more content feature vectors indicative of the current action context. For example, the received data may be used to determine values ​​for certain context parameters, such as the type of location, weather, brightness, congestion, etc., which may be used to construct the context feature vector(s). Any generated context feature vectors, such as those measured by any suitable sensors, such as a camera, and which may include steering patterns, lane keeping patterns, or EEG / ECG data, among others, as well as any user profile vectors (if available), and a set of behavioral parameters, may be provided as inputs to a time network or other state determination network or module 556. This time network may be used to estimate a state value, such as a drowsiness level, for the person based at least in part on this input 558. The context feature vectors and / or user profile data (which may also be provided in the form of one or more feature vectors) may be used to attempt to normalize the behavioral parameters to account for behavioral variations that may not be due to changes in the person's state, but instead may be due to changes in the current action context or person-specific behavioral patterns. The estimated state value generated by this network may then be provided as a measure of the person's current state 560.

[0063] Once a state estimate is generated, information about the estimate may be used for various purposes. For example, a vehicle driver's drowsiness state may have a value or range of values ​​at which drowsiness occurs that causes some action to be taken, such as notifying the user, sounding an alarm, or displaying an icon indicating that the user is detected to be becoming drowsy. In some examples, the message or alarm may be customized for the user or driving context, such that the action relevant to someone on a city highway on a sunny day may be different from someone driving on a rural road during a snowstorm. Furthermore, different users may be likely to take different actions or may not accept some recommendations (e.g., to drink coffee or caffeine). In at least some examples, recommendations may be made based at least in part on historical or preference data about the user, such as if the user previously stopped at a coffee shop during a long drive or if the user indicated that they do not drink caffeine. There may be other drowsiness levels or ranges that cause specific actions to be taken by the vehicle. This may include, for example, activating a driver assistance process or changing the amount of driver assistance provided. For example, lane keeping and automatic braking control may be increased if the driver is determined to be in a particular state of drowsiness. In some situations, the vehicle's control system 336 may be programmed to take drastic action due to the high state of drowsiness indicated by the output drowsiness estimate 330, such as entering a fully autonomous mode or pulling over until such time as the driver is no longer in a high state of drowsiness. In some cases, there may be a regularity requirement for sounding an alarm or taking a particular action when the driver is determined to be in a certain state. These actions may vary by location, jurisdiction, vehicle, type of subject, type of activity, or other such factors.

[0064] The various techniques presented herein are lightweight enough to run in real time on various types of devices, such as personal computers, smart vehicles, or gaming consoles. Such processing may be performed using data captured or generated on the device or received from an external source, such as streaming data received over at least one network. The source may be any suitable source, such as a separate client device, a streaming data provider, or a third-party data provider, among other options. In some cases, the processing and / or use of this data may be performed by one of these other devices, systems, or entities and then provided to the client device (or another such recipient) for presentation or another such use.

[0065] As an example, FIG. 6 illustrates an exemplary network configuration 600 that may be used to provide, generate, modify, encode, and / or transmit data. In at least one embodiment, a client device 602 may generate or receive data for a session using components of a state monitoring application 604 on the client device 602 and data stored locally on the client device. In at least one embodiment, a state monitoring application 624 (e.g., an image generation or editing application) running on a data or content server 620 (e.g., a cloud server or an edge server) may initiate a session associated with at least the client device 602, which may utilize a session manager and user data stored in a user database 634, causing content to be determined by a content manager 626. An estimator module 628 may attempt to estimate state data for one or more subjects based on the received data and may work with a context module 630 to receive context data determined from the received data. At least a portion of that data or state estimate may then be transmitted to the client device 602 using an appropriate transmission manager 622 for delivery via download, streaming, or another such transmission channel. An encoder may be used to encode and / or compress this data before transmission to the client device 602. In at least one embodiment, this data 632 may include any data related to the state estimate, user behavior, or action context. In at least one embodiment, a client device 602 receiving this data may provide this data to a corresponding state monitor 604, which may also or alternatively include a state estimator 612 or context determination module 614 for analyzing the data received by or captured by the client device 602.A decoder may also be used to decode data received over network(s) 640 for presentation or action via client device 602, such as notification content through display 606 or audio, such as an alarm or audible notification, through at least one audio playback device 608, such as a speaker or headphones. In at least one embodiment, at least some of this data may already be stored on, generated on, or accessible to client device 602, such that transmission over network 640 is not required for at least that portion of the data, such as if the data was previously downloaded or stored locally on a hard drive or optical disk. In at least one embodiment, a transmission mechanism, such as data streaming, may be used to transfer this data from server 620 to client device 602, such as from profile database 632. In at least one embodiment, at least a portion of this data may be obtained, determined, or streamed from another source, such as a separate client device 650 or third-party service 660, which may also include functionality for estimating state, determining user behavior patterns, or determining action context. In at least one embodiment, some of this functionality may be implemented using multiple computing devices or multiple processors within one or more computing devices, such as those that may include a combination of a CPU and a GPU.

[0066] In this example, the client devices may include any suitable computing device, such as a desktop computer, a notebook computer, a set-top box, a streaming device, a gaming console, a smartphone, a tablet computer, a smart vehicle, a robotic-assisted machine, a virtual reality headset, an augmented reality goggles, a wearable computer, or a smart television. Each client device may submit requests over at least one wired or wireless network, such as the Internet, an Ethernet, a local area network (LAN), or a cellular network, among other options. In this example, these requests may be submitted to an address associated with a cloud provider, which may operate or control one or more electronic resources in the cloud provider environment, such as a data center or server farm. In at least one embodiment, the requests may be received or processed by at least one edge server located on the network edge and outside at least one security layer associated with the cloud provider environment. In this way, latency may be reduced by allowing client devices to interact with servers that are closer to them, while also improving the security of resources in the cloud provider environment.

[0067] In at least one embodiment, such a system may be used to perform graphical rendering operations. In other embodiments, such a system may be used for other purposes, such as to provide image or video content for testing or validating autonomous machine applications or to perform deep learning operations. In at least one embodiment, such a system may be implemented using edge devices or may incorporate one or more virtual machines (VMs). In at least one embodiment, such a system may be implemented at least in part in a data center or at least in part using cloud computing resources.

[0068] Inference and Training Logic 7A illustrates inference and / or training logic 715 used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with FIG. 7A and / or FIG. 7B.

[0069] In at least one embodiment, the inference and / or training logic 715 may include, but is not limited to, code and / or data storage 701 for storing forward and / or output weights and / or input / output data and / or other parameters for configuring neurons or layers of a neural network to be trained and / or used for inference in one or more embodiments. In at least one embodiment, the training logic 715 may include or be coupled to code and / or data storage 701 for storing graph code or other software for controlling the timing and / or order in which the weights and / or other parameter information should be loaded to configure logic including integer and / or floating-point units (collectively, arithmetic logic units (ALUs)). In at least one embodiment, code such as graph code loads the weights or other parameter information into a processor ALU based on the architecture of the neural network to which the code corresponds. In at least one embodiment, code and / or data storage 701 stores weight parameters and / or input / output data for each layer of a neural network that is trained or used in conjunction with one or more embodiments during forward propagation of the input / output data and / or weight parameters during training and / or inference using aspects of one or more embodiments. In at least one embodiment, any portion of code and / or data storage 701 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory.

[0070] In at least one embodiment, any portion of code and / or data storage 701 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, code and / or code and / or data storage 701 may be cache memory, dynamic randomly addressable memory (“DRAM”), static randomly addressable memory (“SRAM”), non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, the selection of whether code and / or code and / or data storage 701 is internal or external to the processor, or whether it consists of DRAM, SRAM, flash, or some other storage type, may depend on available storage, on-chip versus off-chip, latency requirements of the training and / or inference functions being performed, batch size of data used in inferencing and / or training of the neural network, or some combination of these factors.

[0071] In at least one embodiment, the inference and / or training logic 715 may include, but is not limited to, code and / or data storage 705 for storing back and / or output weights and / or input / output data corresponding to neurons or layers of a neural network trained and / or used to infer in accordance with aspects of one or more embodiments. In at least one embodiment, the code and / or data storage 705 stores weight parameters and / or input / output data for each layer of a neural network trained or used in conjunction with one or more embodiments during backpropagation of input / output data and / or weight parameters during training and / or inference using aspects of one or more embodiments. In at least one embodiment, training logic 715 may include or be coupled to code and / or data storage 705 for storing graph code or other software for controlling timing and / or order, where weights and / or other parameter information should be loaded to configure logic including integer and / or floating-point units (collectively, arithmetic logic units (ALUs)). In at least one embodiment, code, such as graph code, loads weights or other parameter information into processor ALUs based on the architecture of the neural network to which the code corresponds. In at least one embodiment, any portion of code and / or data storage 705 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory. In at least one embodiment, any portion of code and / or data storage 705 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, code and / or data storage 705 may be cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other storage.In at least one embodiment, the selection of whether code and / or data storage 705 is internal or external to the processor, for example, or whether it consists of DRAM, SRAM, flash, or some other storage type, may depend on available storage, on-chip versus off-chip, latency requirements of the training and / or inference functions being performed, batch sizes of data used in inferring and / or training of the neural network, or some combination of these factors.

[0072] In at least one embodiment, code and / or data storage 701 and code and / or data storage 705 may be separate storage structures. In at least one embodiment, code and / or data storage 701 and code and / or data storage 705 may be the same storage structure. In at least one embodiment, code and / or data storage 701 and code and / or data storage 705 may be partially the same storage structure and partially separate storage structures. In at least one embodiment, code and / or data storage 701 and any portion of code and / or data storage 705 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory.

[0073] In at least one embodiment, the inference and / or training logic 715 may include one or more arithmetic logic units (“ALUs”) 710, including, but not limited to, integer and / or floating point units, for performing logical and / or mathematical operations based at least in part on or indicated by the training and / or inference code (e.g., graph code), the results of which may produce activations (e.g., output values ​​from layers or neurons in a neural network) that are stored in activation storage 720, where these activations are a function of input / output and / or weight parameter data stored in code and / or data storage 701 and / or code and / or data storage 705. In at least one embodiment, the activations stored in activation storage 720 are generated according to linear algebra and / or matrix-based mathematics performed by ALU(s) 710 in response to executing instructions or other code, and the weight values ​​stored in code and / or data storage 705 and / or code and / or data storage 701 are used as operands along with other values, such as bias values, gradient information, momentum values, or other parameters or hyperparameters, any or all of which may be stored in code and / or data storage 705 or code and / or data storage 701, or in another storage, on-chip or off-chip.

[0074] In at least one embodiment, ALU(s) 710 are contained within one or more processors or other hardware logic devices or circuits, while in other embodiments, ALU(s) 710 may be external to the processor or other hardware logic device or circuit (e.g., a coprocessor) that uses them. In at least one embodiment, ALU 710 may be contained within an execution unit of a processor, or otherwise within a bank of ALUs accessible by execution units of a processor, either within the same processor or distributed among different processors of different types (e.g., central processing units, graphics processing units, fixed function units, etc.). In at least one embodiment, code and / or data storage 701, code and / or data storage 705, and activation storage 720 may be on the same processor or other hardware logic device or circuit, while in other embodiments, they may be in different processors or other hardware logic devices or circuits, or in some combination of the same processor or other hardware logic device or circuit and different processors or other hardware logic devices or circuits. In at least one embodiment, any portion of activation storage 720 may be included with other on-chip or off-chip data storage, including the processor's L1, L2, or L3 cache or system memory. Additionally, inference and / or training code may be stored with other code accessible to the processor or other hardware logic or circuitry, and may be fetched and / or processed using the processor's fetch, decode, schedule, execute, retirement, and / or other logic circuitry.

[0075] In at least one embodiment, activation storage 720 may be cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, activation storage 720 may be completely or partially within or external to one or more processors or other logic circuits. In at least one embodiment, the selection of whether activation storage 720 is internal or external to a processor, or whether it comprises DRAM, SRAM, flash, or some other storage type, may depend on available storage, on-chip versus off-chip, latency requirements of the training and / or inference functions being performed, batch sizes of data used in inference and / or training of neural networks, or some combination of these factors. In at least one embodiment, the inference and / or training logic 715 shown in FIG. 7a may be used in conjunction with an application-specific integrated circuit ("ASIC"), such as a Tensorflow® processing unit from Google, an inference processing unit (IPU) from Graphcore™, or a Nervana® (e.g., "Lake Crest") processor from Intel Corp. In at least one embodiment, the inference and / or training logic 715 shown in FIG. 7a may be used in conjunction with other hardware, such as central processing unit ("CPU") hardware, graphics processing unit ("GPU") hardware, or a field programmable gate array ("FPGA").

[0076] FIG. 7b illustrates inference and / or training logic 715, according to at least one or more embodiments. In at least one embodiment, the inference and / or training logic 715 may include, but is not limited to, hardware logic in which computational resources are dedicated or otherwise used only in conjunction with weight values ​​or other information corresponding to one or more layers of neurons in a neural network. In at least one embodiment, the inference and / or training logic 715 illustrated in FIG. 7b may be used in conjunction with an application-specific integrated circuit (ASIC), such as a Tensorflow® processing unit from Google, an inference processing unit (IPU) from Graphcore™, or a Nervana® (e.g., “Lake Crest”) processor from Intel Corp. In at least one embodiment, the inference and / or training logic 715 illustrated in FIG. 7b may be used in conjunction with central processing unit (CPU) hardware, graphics processing unit (GPU) hardware, or other hardware, such as a field-programmable gate array (FPGA). In at least one embodiment, inference and / or training logic 715 includes, but is not limited to, code and / or data storage 701 and code and / or data storage 705, which may be used to store code (e.g., graph code), weight values, and / or other information, including bias values, gradient information, momentum values, and / or other parameter or hyperparameter information. In at least one embodiment shown in FIG. 7b, code and / or data storage 701 and code and / or data storage 705 are each associated with dedicated computational resources, such as computation hardware 702 and computation hardware 706, respectively. In at least one embodiment, computation hardware 702 and computation hardware 706 each include one or more ALUs that perform mathematical functions, such as linear algebra functions, solely on the information stored in code and / or data storage 701 and code and / or data storage 705, respectively, with the results stored in activation storage 720.

[0077] In at least one embodiment, each of the code and / or data storage 701 and 705 and corresponding computation hardware 702 and 706 corresponds to a different layer of a neural network, whereby activations resulting from one “storage / computation pair 701 / 702” of code and / or data storage 701 and computation hardware 702 are provided as input to a “storage / computation pair 705 / 706” of code and / or data storage 705 and computation hardware 706 to mirror the conceptual organization of the neural network. In at least one embodiment, each of the storage / computation pairs 701 / 702 and 705 / 706 may correspond to two or more neural network layers. In at least one embodiment, additional storage / computation pairs (not shown) may be included in the inference and / or training logic 715 after or in parallel with the storage / computation pairs 701 / 702 and 705 / 706.

[0078] Data Center 8 illustrates an exemplary data center 800 in which at least one embodiment may be used. In at least one embodiment, data center 800 includes a data center infrastructure layer 810, a framework layer 820, a software layer 830, and an application layer 840.

[0079] 8, data center infrastructure layer 810 may include a resource orchestrator 812, grouped computing resources 814, and node computing resources (“node CRs”) 816(1) through 816(N), where “N” represents any positive integer. In at least one embodiment, node CRs 816(1) through 816(N) may include, without limitation, any number of central processing units (“CPUs”) or other processors (including accelerators, field programmable gate arrays (FPGAs), graphics processors, etc.), memory devices (e.g., dynamic read-only memory), storage devices (e.g., solid state or disk drives), network input / output (“NW I / O”) devices, network switches, virtual machines (“VMs”), power modules, cooling modules, etc. In at least one embodiment, one or more of the nodes CR from among the nodes CR 816(1) through 816(N) may be a server having one or more of the computing resources described above.

[0080] In at least one embodiment, the grouped computing resources 814 may include a distinct grouping of node CRs housed within one or more racks (not shown), or many racks housed in a data center at various geographic locations (also not shown). The distinct groupings of node CRs within the grouped computing resources 814 may include grouped compute resources, network resources, memory resources, or storage resources that may be configured or allocated to support one or more workloads. In at least one embodiment, several node CRs, including CPUs or processors, may be grouped within one or more racks to provide compute resources to support one or more workloads. In at least one embodiment, the one or more racks may also include any number of power modules, cooling modules, and network switches, in any combination.

[0081] In at least one embodiment, resource orchestrator 812 may configure or otherwise control one or more nodes CR 816(1)-816(N) and / or grouped computing resources 814. In at least one embodiment, resource orchestrator 812 may include a software design infrastructure (“SDI”) management entity for data center 800. In at least one embodiment, resource orchestrator may include hardware, software, or some combination thereof.

[0082] 8 , framework layer 820 includes a job scheduler 822, a configuration manager 824, a resource manager 826, and a distributed file system 828. In at least one embodiment, framework layer 820 may include a framework for supporting software 832 in software layer 830 and / or one or more applications 842 in application layer 840. In at least one embodiment, software 832 or application(s) 842 may include web-based service software or applications, such as those offered by Amazon Web Services, Google Cloud, and Microsoft Azure, respectively. In at least one embodiment, framework layer 820 may be a type of free and open-source software web application framework, such as, but not limited to, Apache Spark™ (hereinafter “Spark”), which may utilize distributed file system 828 for large-scale data processing (e.g., “big data”). In at least one embodiment, job scheduler 822 may include a Spark driver to facilitate scheduling of workloads supported by various tiers of data center 800. In at least one embodiment, configuration manager 824 may be capable of configuring different tiers, such as software tier 830, as well as framework tier 820 including Spark and distributed file system 828 to support large-scale data processing. In at least one embodiment, resource manager 826 may be capable of managing clustered or grouped computing resources mapped or allocated to support distributed file system 828 and job scheduler 822. In at least one embodiment, the clustered or grouped computing resources may include grouped computing resources 814 in data center infrastructure tier 810.In at least one embodiment, resource manager 826 may manage these mapped or allocated computing resources in coordination with resource orchestrator 812.

[0083] In at least one embodiment, software 832 included in software layer 830 may include software used by nodes CR 816(1)-816(N), grouped computing resources 814, and / or at least a portion of distributed file system 828 of framework layer 820. The one or more types of software may include, but are not limited to, internet web page searching software, email virus scanning software, database software, and streaming video content software.

[0084] In at least one embodiment, the application(s) 842 included in the application layer 840 may include one or more types of applications used by the nodes CR 816(1)-816(N), the grouped computing resources 814, and / or at least a portion of the distributed file system 828 of the framework layer 820. The one or more types of applications may include, but are not limited to, any number of genomics applications, cognitive compute, and machine learning applications, including training or inference software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), or other machine learning applications used in conjunction with one or more embodiments.

[0085] In at least one embodiment, any of configuration manager 824, resource manager 826, and resource orchestrator 812 may implement any number and types of self-correcting actions based on any amount and type of data obtained in any technically feasible manner. In at least one embodiment, the self-correcting actions may relieve data center operators of data center 800 from determining potentially faulty configurations and potentially avoiding underutilized and / or underperforming portions of the data center.

[0086] In at least one embodiment, data center 800 may include tools, services, software, or other resources for training one or more machine learning models or for predicting or inferring information using one or more machine learning models according to one or more embodiments described herein. For example, in at least one embodiment, machine learning models may be trained by calculating weight parameters according to a neural network architecture using the software and computing resources described above with respect to data center 800. In at least one embodiment, the trained machine learning models corresponding to the one or more neural networks may be used to infer or predict information using the resources described above with respect to data center 800 by using the weight parameters calculated through one or more training techniques described herein.

[0087] In at least one embodiment, the data center may use a CPU, application specific integrated circuit (ASIC), GPU, FPGA, or other hardware to perform training and / or inference using the resources described above. Additionally, one or more of the software and / or hardware resources described above may be configured as a service to enable a user to train or perform inference on information, such as image recognition, speech recognition, or other artificial intelligence services.

[0088] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with Figures 7A and / or 7B. In at least one embodiment, inference and / or training logic 715 may be used in the system of Figure 8 for inference or prediction operations based at least in part on weight parameters calculated using neural network training operations, neural network functionality and / or architecture, or neural network use cases described herein.

[0089] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0090] Computer Systems 9 is a block diagram illustrating an exemplary computer system, which may be a system 900 having interconnected devices and components, a system-on-a-chip (SOC), or some combination thereof, formed with a processor that may include an execution unit for executing instructions, according to at least one embodiment. In at least one embodiment, computer system 900 may include components such as processor 902 to employ an execution unit that includes logic for implementing algorithms to process data according to the present disclosure, such as, but not limited to, embodiments described herein. In at least one embodiment, computer system 900 may include a processor such as the PENTIUM® Processor Family, Xeon™, Itanium®, XScale™, and / or StrongARM™, Intel® Core™, or Intel® Nervana™ microprocessors available from Intel Corporation of Santa Clara, California, although other systems (including PCs with other microprocessors, engineering workstations, set-top boxes, etc.) may also be used. In at least one embodiment, computer system 900 may run a version of the WINDOWS® operating system available from Microsoft Corporation of Redmond, Washington, although other operating systems (e.g., UNIX® and Linux®), embedded software, and / or graphical user interfaces may also be used.

[0091] Embodiments may be used in other devices, such as handheld devices and embedded applications. Some examples of handheld devices include cellular phones, Internet Protocol devices, digital cameras, personal digital assistants ("PDAs"), and handheld PCs. In at least one embodiment, an embedded application may include a microcontroller, a digital signal processor ("DSP"), a system-on-chip, a network computer ("NetPC"), a set-top box, a network hub, a wide area network ("WAN") switch, or any other system capable of implementing one or more instructions according to at least one embodiment.

[0092] In at least one embodiment, computer system 900 may include, but is not limited to, a processor 902, which may include one or more execution units 908 for performing machine learning model training and / or inference according to the techniques described herein. In at least one embodiment, computer system 900 is a single-processor desktop or server system, while in other embodiments, computer system 900 may be a multi-processor system. In at least one embodiment, processor 902 may include, but is not limited to, a complex instruction set computer (“CISC”) microprocessor, a reduced instruction set computing (“RISC”) microprocessor, a very long instruction word (“VLIW”) microprocessor, a processor implementing a combination of instruction sets, or any other processor device, such as, for example, a digital signal processor. In at least one embodiment, the processor 902 may be coupled to a processor bus 910 that may transmit data signals between the processor 902 and other components in the computer system 900.

[0093] In at least one embodiment, processor 902 may include, but is not limited to, level 1 ("L1") internal cache memory ("cache") 904. In at least one embodiment, processor 902 may have a single internal cache or multiple levels of internal cache. In at least one embodiment, cache memory may reside external to processor 902. Other embodiments may include a combination of both internal and external cache, depending on the particular implementation and needs. In at least one embodiment, register file 906 may store different types of data in various registers, including, but not limited to, integer registers, floating-point registers, status registers, and instruction pointer registers.

[0094] In at least one embodiment, an execution unit 908, including logic for performing integer and floating-point operations, is also present in the processor 902. In at least one embodiment, the processor 902 may also include a microcode (“u-code”) read-only memory (“ROM”) that stores microcode for some macroinstructions. In at least one embodiment, the execution unit 908 may include logic for dealing with a packed instruction set 909. In at least one embodiment, by including the packed instruction set 909, along with associated circuitry for executing the instructions, in the instruction set of the general-purpose processor 902, operations used by many multimedia applications may be performed using packed data in the general-purpose processor 902. In one or more embodiments, many multimedia applications may be accelerated and run more efficiently by using the full width of the processor's data bus to perform operations on packed data, which may eliminate the need to transfer smaller units of data across the processor's data bus to perform one or more operations one data element at a time.

[0095] In at least one embodiment, the execution unit 908 may also be used in a microcontroller, an embedded processor, a graphics device, a DSP, and other types of logic circuitry. In at least one embodiment, the computer system 900 may include, but is not limited to, a memory 920. In at least one embodiment, the memory 920 may be implemented as a dynamic random access memory ("DRAM") device, a static random access memory ("SRAM") device, a flash memory device, or other memory device. In at least one embodiment, the memory 920 may store instruction(s) 919 and / or data 921 represented by data signals that may be executed by the processor 902.

[0096] In at least one embodiment, a system logic chip may be coupled to the processor bus 910 and the memory 920. In at least one embodiment, the system logic chip may include, but is not limited to, a memory controller hub (“MCH”) 916, and the processor 902 may communicate with the MCH 916 via the processor bus 910. In at least one embodiment, the MCH 916 may provide a high-bandwidth memory path 918 to the memory 920 for instruction and data storage, and for storage of graphics commands, data, and textures. In at least one embodiment, the MCH 916 may direct data signals between the processor 902, the memory 920, and other components in the computer system 900, and may bridge data signals between the processor bus 910, the memory 920, and the system I / O 922. In at least one embodiment, the system logic chip may provide a graphics port for coupling to a graphics controller. In at least one embodiment, the MCH 916 may be coupled to memory 920 through a high-bandwidth memory path 918, and the graphics / video card 912 may be coupled to the MCH 916 through an Accelerated Graphics Port (“AGP”) interconnect 914.

[0097] In at least one embodiment, computer system 900 may use system I / O 922, a proprietary hub interface bus, to couple MCH 916 to I / O controller hub (“ICH”) 930. In at least one embodiment, ICH 930 may provide direct connectivity to several I / O devices via a local I / O bus. In at least one embodiment, the local I / O bus may include, but is not limited to, a high-speed I / O bus for connecting peripherals to memory 920, a chipset, and processor 902. Examples may include, but are not limited to, an audio controller 929, a firmware hub ("flash BIOS") 928, a wireless transceiver 926, data storage 924, a legacy I / O controller 923 including a user input and keyboard interface 925, a serial expansion port 927 such as a Universal Serial Bus ("USB"), and a network controller 934. Data storage 924 may comprise a hard disk drive, a floppy disk drive, a CD-ROM device, a flash memory device, or other mass storage device.

[0098] In at least one embodiment, Figure 9 illustrates a system including interconnected hardware devices or "chips," although in other embodiments, Figure 9 may illustrate an exemplary system-on-a-chip ("SoC"). In at least one embodiment, the devices may be interconnected with proprietary interconnects, standard interconnects (e.g., PCIe), or some combination thereof. In at least one embodiment, one or more components of computer system 900 are interconnected using a compute express link (CXL) interconnect.

[0099] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with Figures 7A and / or 7B. In at least one embodiment, inference and / or training logic 715 may be used in the system of Figure 9 for inference or prediction operations based at least in part on weight parameters calculated using neural network training operations, neural network functionality and / or architecture, or neural network use cases described herein.

[0100] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0101] 10 is a block diagram illustrating an electronic device 1000 for utilizing a processor 1010, according to at least one embodiment. In at least one embodiment, the electronic device 1000 may be, for example, but not limited to, a notebook, a tower server, a rack server, a blade server, a laptop, a desktop, a tablet, a mobile device, a phone, an embedded computer, or any other suitable electronic device.

[0102] In at least one embodiment, system 1000 may include a processor 1010 communicatively coupled to any suitable number or type of components, peripherals, modules, or devices, including, but not limited to, a 1010 bus, a System Management Bus (“SMBus”), a Low Pin Count (“LPC”) bus, a Serial Peripheral Interface (“SPI”), a High Definition Audio (“HDA”) bus, a Serial Advance Technology Attachment (“SATA”) bus, a Universal Serial Bus (“USB”) (versions 1, 2, and 3), or a Universal Asynchronous Receiver / Transmitter (“UART”) bus. In at least one embodiment, FIG. 10 illustrates a system including interconnected hardware devices or "chips," although in other embodiments, FIG. 10 may illustrate an exemplary system-on-a-chip ("SoC"). In at least one embodiment, the devices illustrated in FIG. 10 may be interconnected with a proprietary interconnect, a standard interconnect (e.g., PCIe), or some combination thereof. In at least one embodiment, one or more components of FIG. 10 are interconnected using a Compute Express Link (CXL) interconnect.

[0103] In at least one embodiment, FIG. 10 includes a display 1024, a touch screen 1025, a touch pad 1030, a Near Field Communication ("NFC") unit 1045, a sensor hub 1040, a thermal sensor 1046, an Express Chipset ("EC") 1035, a Trusted Platform Module ("TPM") 1038, a BIOS / firmware / flash memory ("BIOS,FW flash") 1022, a DSP 1060, a drive 1020, such as a solid state disk ("SSD") or hard disk drive ("HDD"), a wireless local area network ("WLAN") unit 1050, a Bluetooth unit 1052, a wireless wide area network ("WWAN") unit 1054, a Bluetooth module 1056, a Bluetooth-enabled device 1058 ... Network) 1056, Global Positioning System (GPS) 1055, a camera such as a USB 3.0 camera (“USB 3.0 Camera”) 1054, and / or a Low Power Double Data Rate (“LPDDR”) memory unit (“LPDDR3”) implemented, for example, in the LPDDR3 standard. Each of these components may be implemented in any suitable manner.

[0104] In at least one embodiment, other components may be communicatively coupled to the processor 1010 through the components described above. In at least one embodiment, an accelerometer 1041, an ambient light sensor (“ALS”) 1042, a compass 1043, and a gyroscope 1044 may be communicatively coupled to the sensor hub 1040. In at least one embodiment, a thermal sensor 1039, a fan 1037, a keyboard 1046, and a touchpad 1030 may be communicatively coupled to the EC 1035. In at least one embodiment, a speaker 1063, headphones 1064, and a microphone (“mic”) 1065 may be communicatively coupled to an audio unit (“audio codec and class D amplifier”) 1062, which may be communicatively coupled to the DSP 1060. In at least one embodiment, the audio unit 1064 may include, for example, without limitation, an audio coder / decoder ("codec") and a Class D amplifier. In at least one embodiment, a SIM card ("SIM") 1057 may be communicatively coupled to the WWAN unit 1056. In at least one embodiment, components such as the WLAN unit 1050 and Bluetooth unit 1052, and the WWAN unit 1056 may be implemented in a Next Generation Form Factor ("NGFF").

[0105] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with Figures 7a and / or 7b. In at least one embodiment, inference and / or training logic 715 may be used in the system of Figure 10 for inference or prediction operations based at least in part on weight parameters calculated using neural network training operations, neural network functionality and / or architecture, or neural network use cases described herein.

[0106] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0107] 11 is a block diagram of a processing system according to at least one embodiment. In at least one embodiment, system 1100 includes one or more processors 1102 and one or more graphics processors 1108 and may be a single-processor desktop system, a multiprocessor workstation system, or a server system with multiple processors 1102 or processor cores 1107. In at least one embodiment, system 1100 is a processing platform integrated into a system-on-chip (SoC) integrated circuit for use in a mobile, handheld, or embedded device.

[0108] In at least one embodiment, system 1100 may include or be incorporated within a server-based gaming platform, a game console including a game and media console, a mobile gaming console, a handheld game console, or an online game console. In at least one embodiment, system 1100 is a mobile phone, a smart phone, a tablet computing device, or a mobile internet device. In at least one embodiment, processing system 1100 may also include, be coupled to, or be incorporated within a wearable device, such as a smart watch wearable device, a smart eyewear device, an augmented reality device, or a virtual reality device. In at least one embodiment, processing system 1100 is a television or set-top box device having one or more processors 1102 and a graphical interface generated by one or more graphics processors 1108.

[0109] In at least one embodiment, the one or more processors 1102 each include one or more processor cores 1107 for processing instructions that, when executed, perform operations for system and user software. In at least one embodiment, each of the one or more processor cores 1107 is configured to process a particular instruction set 1109. In at least one embodiment, the instruction set 1109 may facilitate Complex Instruction Set Computing (CISC), Reduced Instruction Set Computing (RISC), or computing via Very Long Instruction Word (VLIW). In at least one embodiment, the processor cores 1107 may each process a different instruction set 1109, and the instruction set 1109 may include instructions to facilitate emulation of other instruction sets. In at least one embodiment, the processor cores 1107 may also include other processing devices, such as a digital signal processor (DSP).

[0110] In at least one embodiment, the processor 1102 includes cache memory 1104. In at least one embodiment, the processor 1102 can have a single internal cache or multiple levels of internal cache. In at least one embodiment, the cache memory is shared among various components of the processor 1102. In at least one embodiment, the processor 1102 also uses an external cache (e.g., a level 3 (L3) cache or a last level cache (LLC)) (not shown), which can be shared among the processor cores 1107 using known cache coherency techniques. In at least one embodiment, a register file 1106 is additionally included in the processor 1102, which can include different types of registers (e.g., integer registers, floating-point registers, status registers, and instruction pointer registers) for storing different types of data. In at least one embodiment, the register file 1106 can include general-purpose registers or other registers.

[0111] In at least one embodiment, one or more processors 1102 are coupled to one or more interface buses 1110 for transmitting communication signals, such as address, data, or control signals, between the processors 1102 and other components in the system 1100. In at least one embodiment, the interface bus 1110 in one embodiment may be a processor bus, such as a version of a Direct Media Interface (DMI) bus. In at least one embodiment, the interface 1110 is not limited to a DMI bus, but may include one or more peripheral component interconnect buses (e.g., PCI, PCI Express), a memory bus, or other types of interface buses. In at least one embodiment, the processor(s) 1102 include an integrated memory controller 1116 and a platform controller hub 1130. In at least one embodiment, a memory controller 1116 facilitates communication between memory devices and other components of the system 1100, and a platform controller hub (PCH) 1130 provides connectivity to I / O devices via a local I / O bus.

[0112] In at least one embodiment, memory device 1120 may be a dynamic random access memory (DRAM) device, a static random access memory (SRAM) device, a flash memory device, a phase change memory device, or any other memory device with performance suitable for serving as process memory. In at least one embodiment, memory device 1120 may operate as system memory for system 1100 to store data 1122 and instructions 1121 for use when one or more processors 1102 execute applications or processes. In at least one embodiment, memory controller 1116 also couples to an optional external graphics processor 1112, which may communicate with one or more graphics processors 1108 in processor 1102 to perform graphics and media operations. In at least one embodiment, display device 1111 may be connected to processor(s) 1102. In at least one embodiment, display device 1111 may include one or more of an internal display device, such as in a mobile electronic device or laptop device, or an external display device attached via a display interface (e.g., DisplayPort, etc.). In at least one embodiment, display device 1111 may include a head mounted display (HMD), such as a stereoscopic display device for use in virtual reality (VR) or augmented reality (AR) applications.

[0113] In at least one embodiment, the platform controller hub 1130 allows peripherals to connect to the memory device 1120 and the processor 1102 via a high-speed I / O bus. In at least one embodiment, the I / O peripherals include, but are not limited to, an audio controller 1146, a network controller 1134, a firmware interface 1128, a wireless transceiver 1126, a touch sensor 1125, and a data storage device 1124 (e.g., a hard disk drive, flash memory, etc.). In at least one embodiment, the data storage device 1124 can be connected via a storage interface (e.g., SATA) or via a peripheral bus such as a peripheral component interconnect bus (e.g., PCI, PCI Express). In at least one embodiment, the touch sensor 1125 can include a touch screen sensor, a pressure sensor, or a fingerprint sensor. In at least one embodiment, wireless transceiver 1126 may be a Wi-Fi transceiver, a Bluetooth transceiver, or a mobile network transceiver such as a 3G, 4G, or Long Term Evolution (LTE) transceiver. In at least one embodiment, firmware interface 1128 enables communication with system firmware and may be, for example, a unified extensible firmware interface (UEFI). In at least one embodiment, network controller 1134 may enable network connectivity to a wired network. In at least one embodiment, a high-performance network controller (not shown) couples to interface bus 1110. In at least one embodiment, audio controller 1146 is a multi-channel high-definition audio controller.In at least one embodiment, system 1100 includes an optional legacy I / O controller 1140 for coupling legacy (e.g., Personal System 2 (PS / 2)) devices to the system. In at least one embodiment, platform controller hub 1130 can also connect to one or more universal serial bus (USB) controller 1142-connected input devices, such as a keyboard and mouse 1143 combination, a camera 1144, or other USB input devices.

[0114] In at least one embodiment, instances of memory controller 1116 and platform controller hub 1130 may be embedded in a discreet external graphics processor, such as external graphics processor 1112. In at least one embodiment, platform controller hub 1130 and / or memory controller 1116 may be external to one or more processors 1102. For example, in at least one embodiment, system 1100 may include external memory controller 1116 and platform controller hub 1130, which may be configured as a memory controller hub and a peripheral controller hub within a system chipset in communication with processor(s) 1102.

[0115] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with FIG. 7A and / or FIG. 7B . In at least one embodiment, portions or all of inference and / or training logic 715 may be incorporated into graphics processor 1500. For example, in at least one embodiment, the training and / or inference techniques described herein may use one or more of the ALUs embodied in the graphics processor. Moreover, in at least one embodiment, the inference and / or training operations described herein may be performed using logic other than that shown in FIG. 7A or FIG. 7B . In at least one embodiment, the weight parameters may be stored in on-chip or off-chip memory and / or registers (shown or not shown) that configure the ALUs of the graphics processor for implementing one or more machine learning algorithms, neural network architectures, use cases, or training techniques described herein.

[0116] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0117] 12 is a block diagram of a processor 1200 having one or more processor cores 1202A-1202N, an integrated memory controller 1214, and an integrated graphics processor 1208, according to at least one embodiment. In at least one embodiment, the processor 1200 may include additional cores, up to additional core 1202N, represented by dashed boxes. In at least one embodiment, each of the processor cores 1202A-1202N includes one or more internal cache units 1204A-1204N. In at least one embodiment, each processor core also has access to one or more shared cached units 1206.

[0118] In at least one embodiment, the internal cache units 1204A-1204N and the shared cache unit 1206 represent a cache memory hierarchy within the processor 1200. In at least one embodiment, the cache memory units 1204A-1204N may include at least one level of instruction and data cache within each processor core, and one or more levels of shared mid-level cache, such as level 2 (L2), level 3 (L3), level 4 (L4), or other levels of cache, where the highest level of cache before external memory is classified as LLC. In at least one embodiment, cache coherency logic maintains coherency between the various cache units 1206 and 1204A-1204N.

[0119] In at least one embodiment, processor 1200 may also include a set of one or more bus controller units 1216 and a system agent core 1210. In at least one embodiment, one or more bus controller units 1216 manage a set of peripheral buses, such as one or more PCI or PCI Express buses. In at least one embodiment, system agent core 1210 provides management functionality for various processor components. In at least one embodiment, system agent core 1210 includes one or more integrated memory controllers 1214 for managing access to various external memory devices (not shown).

[0120] In at least one embodiment, one or more of the processor cores 1202A-1202N include support for simultaneous multithreading. In at least one embodiment, the system agent core 1210 includes components for coordinating and operating the cores 1202A-1202N during multithreaded processing. In at least one embodiment, the system agent core 1210 may additionally include a power control unit (PCU), which includes logic and components for adjusting the power state of one or more of the processor cores 1202A-1202N and the graphics processor 1208.

[0121] In at least one embodiment, processor 1200 additionally includes a graphics processor 1208 for performing graphics processing operations. In at least one embodiment, graphics processor 1208 couples to a system agent core 1210 that includes a shared cache unit 1206 and one or more integrated memory controllers 1214. In at least one embodiment, system agent core 1210 also includes a display controller 1211 for driving graphics processor output to one or more coupled displays. In at least one embodiment, display controller 1211 may also be a separate module coupled to graphics processor 1208 via at least one interconnect or may be incorporated within graphics processor 1208.

[0122] In at least one embodiment, a ring-based interconnect unit 1212 is used to couple the internal components of processor 1200. In at least one embodiment, alternative interconnect units, such as a point-to-point interconnect, a switched interconnect, or other techniques, may be used. In at least one embodiment, graphics processor 1208 couples to ring interconnect 1212 via I / O link 1213.

[0123] In at least one embodiment, I / O link 1213 represents at least one of several types of I / O interconnect, including an on-package I / O interconnect that facilitates communication between various processor components and a high-performance embedded memory module 1218, such as an eDRAM module. In at least one embodiment, each of processor cores 1202A-1202N and graphics processor 1208 use embedded memory module 1218 as a shared last-level cache.

[0124] In at least one embodiment, processor cores 1202A-1202N are homogeneous cores that execute a common instruction set architecture. In at least one embodiment, processor cores 1202A-1202N are heterogeneous in terms of instruction set architecture (ISA), where one or more of processor cores 1202A-1202N execute a common instruction set and one or more other of processor cores 1202A-1202N execute a subset of the common instruction set or a different instruction set. In at least one embodiment, processor cores 1202A-1202N are heterogeneous in terms of microarchitecture, where one or more cores with relatively high power consumption are combined with one or more cores with lower power consumption. In at least one embodiment, processor 1200 may be implemented on one or more chips or as an SoC integrated circuit.

[0125] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding the inference and / or training logic 715 are provided below in conjunction with FIG. 7a and / or FIG. 7b. In at least one embodiment, portions or all of the inference and / or training logic 715 may be incorporated into processor 1200. For example, in at least one embodiment, the training and / or inference techniques described herein may use one or more of the ALUs embodied in graphics processor 1512, graphics core(s) 1202A-1202N, or other components in FIG. 12. Moreover, in at least one embodiment, the inference and / or training operations described herein may be performed using logic other than that shown in FIG. 7A or FIG. 7B. In at least one embodiment, the weight parameters may be stored in on-chip or off-chip memory and / or registers (shown or not shown) that comprise the ALU of graphics processor 1200 for implementing one or more machine learning algorithms, neural network architectures, use cases, or training techniques described herein.

[0126] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0127] Virtualized Computing Platform 13 is an exemplary data flow diagram for a process 1300 for generating and deploying an image processing and inference pipeline, according to at least one embodiment. In at least one embodiment, the process 1300 may be deployed for use with imaging devices, processing devices, and / or other device types at one or more facilities 1302. The process 1300 may be executed within a training system 1304 and / or a deployment system 1306. In at least one embodiment, the training system 1304 may be used to train, deploy, and implement machine learning models (e.g., neural networks, object detection algorithms, computer vision algorithms, etc.) for use in the deployment system 1306. In at least one embodiment, the deployment system 1306 may be configured to offload processing and compute resources between distributed computing environments to reduce infrastructure requirements at the facility 1302. In at least one embodiment, one or more applications in the pipeline may use or call services (e.g., inference, visualization, compute, AI, etc.) of the deployment system 1306 during execution of the application.

[0128] In at least one embodiment, some of the applications used in the advanced processing and inference pipeline may use machine learning models or other AI to perform one or more processing steps. In at least one embodiment, the machine learning models may be trained at facility 1302 using data (e.g., imaging data) 1308 generated at facility 1302 (and stored in one or more picture archiving and communication system (PACS) servers at facility 1302), may be trained using imaging or sequencing data 1308 from another facility(ies), or a combination thereof. In at least one embodiment, training system 1304 may be used to provide applications, services, and / or other resources for generating practical, deployable machine learning models for deployment system 1306.

[0129] In at least one embodiment, model registry 1324 may be backed by object storage that may support versioning and object metadata. In at least one embodiment, object storage may be accessible, for example, from within a cloud platform, through a cloud storage (e.g., cloud 1426 of FIG. 14 ) compatible application programming interface (API). In at least one embodiment, machine learning models in model registry 1324 may be uploaded, listed, modified, or deleted by a system developer or partner interacting with the API. In at least one embodiment, the API may provide access to methods that allow a user with appropriate credentials to associate a model with an application, so that the model can be executed as part of the execution of a containerized instantiation of the application.

[0130] In at least one embodiment, the training pipeline 1404 ( FIG. 14 ) may include a scenario in which the facility 1302 is training its own machine learning models or has existing machine learning models that need to be optimized or updated. In at least one embodiment, imaging data 1308 generated by one or more imaging devices, sequencing devices, and / or other device types may be received. In at least one embodiment, once the imaging data 1308 is received, AI-assisted annotation 1310 may be used to assist in generating annotations corresponding to the imaging data 1308 to be used as ground truth data for the machine learning models. In at least one embodiment, the AI-assisted annotation 1310 may include one or more machine learning models (e.g., convolutional neural networks (CNNs)), which may be trained to generate annotations corresponding to several types of imaging data 1308 (e.g., from several devices). In at least one embodiment, the AI-assisted annotations 1310 may then be used directly or may be adjusted or fine-tuned using an annotation tool to generate ground truth data. In at least one embodiment, the AI-assisted annotations 1310, the labeled clinical data 1312, or a combination thereof may be used as ground truth data for training a machine learning model. In at least one embodiment, the trained machine learning model may be referred to as an output model 1316 and may be used by the deployment system 1306 as described herein.

[0131] In at least one embodiment, the training pipeline 1404 ( FIG. 14 ) may include a scenario in which the facility 1302 needs a machine learning model for use in performing one or more processing tasks for one or more applications in the installation system 1306, but the facility 1302 may not currently have such a machine learning model (or may not have an optimized, efficient, or effective model for such purpose). In at least one embodiment, an existing machine learning model may be selected from the model registry 1324. In at least one embodiment, the model registry 1324 may include machine learning models trained to perform a variety of different inference tasks on imaging data. In at least one embodiment, the machine learning models in the model registry 1324 may have been trained on imaging data from a facility different from the facility 1302 (e.g., a remote facility). In at least one embodiment, the machine learning models may have been trained on imaging data from one location, two locations, or any number of locations. In at least one embodiment, when training on imaging data from a particular location, the training may occur at that location, or at least in a manner that protects the confidentiality of the imaging data or limits its transfer off-premise. In at least one embodiment, once a model is trained or partially trained at one location, the machine learning model may be added to model registry 1324. In at least one embodiment, the machine learning model may then be retrained or updated at any number of other facilities, and the retrained or updated model may be made available in model registry 1324. In at least one embodiment, a machine learning model may then be selected from model registry 1324, sometimes referred to as output model 1316, and used in installation system 1306 to perform one or more processing tasks for one or more applications of the installation system.

[0132] In at least one embodiment, training pipeline 1404 (FIG. 14), a scenario may include facility 1302 needing a machine learning model for use in performing one or more processing tasks for one or more applications in deployment system 1306, but facility 1302 may not currently have such a machine learning model (or may not have a model optimized, efficient, or effective for such purpose). In at least one embodiment, the machine learning model selected from model registry 1324 may not be fine-tuned or optimized for the imaging data 1308 generated at facility 1302 due to differences in the population, robustness of the training data used to train the machine learning model, the variety of anomalies in the training data, and / or other issues with the training data. In at least one embodiment, AI-assisted annotation 1310 may be used to assist in generating annotations corresponding to imaging data 1308 to be used as ground truth data for retraining or updating the machine learning model. In at least one embodiment, labeled data 1312 may be used as ground truth data for training the machine learning model. In at least one embodiment, retraining or updating the machine learning model may be referred to as model training 1314. In at least one embodiment, model training 1314, e.g., AI-assisted annotations 1310, labeled clinic data 1312, or a combination thereof, may be used as ground truth data for retraining or updating the machine learning model. In at least one embodiment, the trained machine learning model may be referred to as output model 1316 and may be used by deployment system 1306 as described herein.

[0133] In at least one embodiment, deployment system 1306 may include software 1318, services 1320, hardware 1322, and / or other components, features, and functionality. In at least one embodiment, deployment system 1306 may include a software “stack” whereby software 1318 may build on and use services 1320 to perform some or all of the processing tasks, and services 1320 and software 1318 may build on and use hardware 1322 to perform the processing, storage, and / or other compute tasks of deployment system 1306. In at least one embodiment, software 1318 may include any number of different containers, each of which may perform an instantiation of an application. In at least one embodiment, each application may perform one or more processing tasks (e.g., inference, object detection, feature detection, segmentation, image enhancement, calibration, etc.) in an advanced processing and inference pipeline. In at least one embodiment, an advanced processing and inference pipeline may be defined based on a selection of different containers desired or needed to process imaging data 1308, in addition to the containers that receive and configure the imaging data for use by each container and / or by facility 1302 after processing through the pipeline (e.g., to convert the output to a usable data type). In at least one embodiment, the combination of containers in software 1318 (e.g., making up a pipeline) may be referred to as a virtual device (as described in more detail herein), which may utilize services 1320 and hardware 1322 to perform some or all processing tasks of applications instantiated in the containers.

[0134] In at least one embodiment, the data processing pipeline may receive input data (e.g., imaging data 1308) in a particular format in response to an inference request (e.g., a request from a user of the deployment system 1306). In at least one embodiment, the input data may represent one or more images, videos, and / or other data representations generated by one or more imaging devices. In at least one embodiment, the data may undergo pre-processing as part of the data processing pipeline to prepare the data for processing by one or more applications. In at least one embodiment, post-processing may be performed on the output of one or more inference tasks or other processing tasks of the pipeline to prepare output data for a subsequent application and / or to prepare output data for transmission and / or use by a user (e.g., in response to an inference request). In at least one embodiment, the inference tasks may be performed by one or more machine learning models, such as trained or deployed neural networks, which may include output model 1316 of the training system 1304.

[0135] In at least one embodiment, tasks in a data processing pipeline may be encapsulated in container(s), with each container(s) representing a separate, fully functional instantiation of an application and a virtualized computing environment capable of referencing machine learning models. In at least one embodiment, containers or applications may be published to a private (e.g., restricted access) area of ​​a container registry (described in more detail herein), and trained or deployed models may be stored in a model registry 1324 and associated with one or more applications. In at least one embodiment, images of applications (e.g., container images) may be available in the container registry, and when selected from the container registry by a user for deployment in the pipeline, the images may be used to generate a container for instantiation of the application for use by the user's system.

[0136] In at least one embodiment, a developer (e.g., a software developer, clinician, physician, etc.) may develop, publish, and store (e.g., as a container) an application for performing image processing and / or inference on the provided data. In at least one embodiment, the development, publishing, and / or storage may be performed using a software development kit (SDK) associated with the system (e.g., to ensure that the developed application and / or container conforms to or is compatible with the system). In at least one embodiment, the developed application may be tested locally (e.g., at a first facility against data from the first facility) using an SDK that may support at least some of the services 1320 as a system (e.g., system 1400 of FIG. 14 ). In at least one embodiment, because a DICOM object may contain anywhere from one to hundreds of images or other data types, and because of data variability, the developer may be responsible for managing the extraction and preparation of the incoming data (e.g., setting up constructs for the application, incorporating pre-processing into the application, etc.). In at least one embodiment, once verified (e.g., for accuracy) by system 1400, the application may be made available in a container registry for selection and / or implementation by a user to perform one or more processing tasks on the data at the user's facility (e.g., a second facility).

[0137] In at least one embodiment, the developer may then share the application or container over a network for access and use by users of the system (e.g., system 1400 of FIG. 14 ). In at least one embodiment, the completed and validated application or container may be stored in a container registry, and the associated machine learning model may be stored in a model registry 1324. In at least one embodiment, a requesting entity providing an inference or image processing request may browse the container registry and / or the model registry 1324 for applications, containers, datasets, machine learning models, etc., select a desired combination of elements for inclusion in a data processing pipeline, and submit the image processing request. In at least one embodiment, the request may include input data (and, in some instances, associated patient data) required to perform the request and / or may include a selection of application(s) and / or machine learning model(s) to be executed in processing the request. In at least one embodiment, the request may then be passed to one or more components of the deployment system 1306 (e.g., the cloud) to perform the processing of the data processing pipeline. In at least one embodiment, processing by the deployment system 1306 may include referencing selected elements (e.g., applications, containers, models, etc.) from the container registry and / or the model registry 1324. In at least one embodiment, once results are produced by the pipeline, the results may be returned to the user for viewing (e.g., for viewing in a viewing application suite running on a local, on-premise workstation or terminal).

[0138] In at least one embodiment, services 1320 may be utilized to assist in the processing or execution of applications or containers in a pipeline. In at least one embodiment, services 1320 may include compute services, artificial intelligence (AI) services, visualization services, and / or other service types. In at least one embodiment, services 1320 may provide functionality that is common to one or more applications in software 1318, and thus the functionality may be abstracted into services that can be called or utilized by the applications. In at least one embodiment, the functionality provided by services 1320 may operate dynamically and more efficiently and may scale well by allowing applications to process data in parallel (e.g., using parallel computing platform 1430 (FIG. 14)). In at least one embodiment, services 1320 may be shared among and between various applications, rather than each application that shares the same functionality provided by service 1320 being required to have its own instance of service 1320. In at least one embodiment, the services may include, by way of non-limiting example, an inference server or engine that may be used to perform detection or segmentation tasks. In at least one embodiment, a model training service may be included that may provide machine learning model training and / or retraining capabilities. In at least one embodiment, a data augmentation service may further be included that may provide GPU-accelerated data (e.g., DICOM, RIS, CIS, REST-compliant, RPC, raw, etc.) extraction, resizing, scaling, and / or other enhancements. In at least one embodiment, a visualization service may be used that may add image rendering effects such as ray tracing, rasterization, denoising, sharpening, etc. to add realism to two-dimensional (2D) and / or three-dimensional (3D) models.In at least one embodiment, virtual device services may be included that provide beamforming, segmentation, inference, imaging, and / or support for other applications in the virtual device pipeline.

[0139] In at least one embodiment, if services 1320 include an AI service (e.g., an inference service), the one or more machine learning models may be executed by calling (e.g., as an API call) an inference service (e.g., an inference server) to execute the machine learning model(s), or processing thereof, as part of the application execution. In at least one embodiment, if another application includes one or more machine learning models for a segmentation task, the application may call the inference service to execute the machine learning models to perform one or more of the processing operations associated with the segmentation task. In at least one embodiment, software 1318 implementing an advanced processing and inference pipeline, including a segmentation application and an anomaly detection application, may be streamlined because each application may call the same inference service to perform one or more inference tasks.

[0140] In at least one embodiment, hardware 1322 may include a GPU, a CPU, a graphics card, an AI / deep learning system (e.g., an AI supercomputer such as NVIDIA's DGX), a cloud platform, or a combination thereof. In at least one embodiment, different types of hardware 1322 may be used to provide efficient and dedicated support for software 1318 and services 1320 in deployment system 1306. In at least one embodiment, the use of GPU processing for processing locally (e.g., at facility 1302), within the AI / deep learning system, in a cloud system, and / or in other processing components of deployment system 1306 may be implemented to improve the efficiency, accuracy, and effectiveness of image processing and generation. In at least one embodiment, software 1318 and / or services 1320 may be optimized for GPU processing for, by way of non-limiting example, deep learning, machine learning, and / or high-performance computing. In at least one embodiment, at least a portion of the computing environment of deployment system 1306 and / or training system 1304 may be executed on one or more supercomputers or high-performance computing systems in a data center using GPU-optimized software (e.g., a hardware and software combination of an NVIDIA DGX system). In at least one embodiment, hardware 1322 may include any number of GPUs, which may be called upon to perform data parallel processing as described herein. In at least one embodiment, the cloud platform may further include GPU processing for GPU-optimized execution of deep learning tasks, machine learning tasks, or other computing tasks. In at least one embodiment, the cloud platform (e.g., NVIDIA NGC) may be executed using one or more AI / deep learning supercomputers and / or GPU-optimized software (e.g., provided on an NVIDIA DGX system) as a hardware abstraction and scaling platform.In at least one embodiment, the cloud platform may integrate an application container clustering system or orchestration system (e.g., Kubernetes) across multiple GPUs to enable seamless scaling and load balancing.

[0141] 14 is a system diagram for an exemplary system 1400 for generating and deploying an imaging deployment pipeline, according to at least one embodiment. In at least one embodiment, system 1400 may be used to implement process 1300 of FIG. 13 and / or other processes, including advanced processing and inference pipelines. In at least one embodiment, system 1400 may include a training system 1304 and a deployment system 1306. In at least one embodiment, training system 1304 and deployment system 1306 may be implemented using software 1318, services 1320, and / or hardware 1322, as described herein.

[0142] In at least one embodiment, system 1400 (e.g., training system 1304 and / or deployment system 1306) may be implemented in a cloud computing environment (e.g., using cloud 1426). In at least one embodiment, system 1400 may be implemented locally with respect to a healthcare service facility or as a combination of both cloud and local computing resources. In at least one embodiment, access to APIs in cloud 1426 may be limited to authorized users through enacted security measures or protocols. In at least one embodiment, the security protocol may include web tokens, which may be signed by an authentication (e.g., AuthN, AuthZ, Gluecon, etc.) service and may have appropriate permissions. In at least one embodiment, the APIs of a virtual appliance (as described herein) or other instantiation of system 1400 may be limited to a set of public IPs that are verified or authorized for interaction.

[0143] In at least one embodiment, the various components of system 1400 may communicate among themselves via wired and / or wireless communication protocols using any of a variety of different network types, including, but not limited to, a local area network (LAN) and / or a wide area network (WAN). In at least one embodiment, communications between the facility and the components of system 1400 (e.g., for sending inference requests, receiving results of inference requests, etc.) may be communicated via data bus(es), wireless data protocol (Wi-Fi), wired data protocol (e.g., Ethernet), etc.

[0144] In at least one embodiment, the training system 1304 may execute a training pipeline 1404 similar to that described herein with respect to FIG. 13 . In at least one embodiment, if one or more machine learning models are to be used by the deployment system 1306 in the deployment pipeline 1410, the training pipeline 1404 may be used to train or retrain one or more (e.g., pre-trained) models and / or implement one or more of the pre-trained models 1406 (e.g., without the need for retraining or updating). In at least one embodiment, the training pipeline 1404 may result in the generation of output model(s) 1316. In at least one embodiment, the training pipeline 1404 may include any number of processing steps, such as, but not limited to, imaging data (or other input data) conversion or adaptation. In at least one embodiment, different training pipelines 1404 may be used for different machine learning models used by the deployment system 1306. In at least one embodiment, a training pipeline 1404 similar to the first example described with respect to FIG. 13 may be used for the first machine learning model, a training pipeline 1404 similar to the second example described with respect to FIG. 13 may be used for the second machine learning model, and a training pipeline 1404 similar to the third example described with respect to FIG. 13 may be used for the third machine learning model. In at least one embodiment, any combination of tasks in training system 1304 may be used depending on what is required for each respective machine learning model. In at least one embodiment, one or more of the machine learning models may already be trained and ready for deployment; thus, the machine learning models may not undergo any processing by training system 1304 and may be implemented by deployment system 1306.

[0145] In at least one embodiment, the output model(s) 1316 and / or the pre-trained model(s) 1406 may include any type of machine learning model, depending on the implementation or embodiment. In at least one embodiment, and without limitation, the machine learning models used by system 1400 may include machine learning model(s) using linear regression, logistic regression, decision trees, support vector machines (SVMs), naive Bayes, k-nearest neighbor (Knn), k-means clustering, random forests, dimensionality reduction algorithms, gradient boosting algorithms, neural networks (e.g., autoencoder, convolutional, recurrent, perceptron, long / short-term memory (LSTM), Hopfield, Boltzmann, deep belief, deconvolution, generative adversarial, liquid state machine, etc.), and / or other types of machine learning models.

[0146] In at least one embodiment, the training pipeline 1404 may include AI-assisted annotation, as described in more detail herein with respect to at least FIG. 15B. In at least one embodiment, the labeled data 1312 (e.g., traditional annotations) may be generated by any number of techniques. In at least one embodiment, the labels or other annotations may be generated within a drawing program (e.g., an annotation program), a computer-aided design (CAD) program, a labeling program, another type of program suitable for generating annotations or labels for ground truth, and / or may be drawn by hand in some instances. In at least one embodiment, the ground truth data can be synthetically generated (e.g., generated from a computer model or rendering), realistically generated (e.g., designed and generated from real-world data), machine-automated (e.g., using feature analysis and learning to extract features from data and then generate labels), human-annotated (e.g., a labeler or annotation expert defines label locations), and / or a combination thereof. In at least one embodiment, for each instance of imaging data 1308 (or other data type used by a machine learning model), there can be corresponding ground truth data generated by training system 1304. In at least one embodiment, AI-assisted annotation can be performed as part of the deployment pipeline 1410, either in addition to or instead of AI-assisted annotation included in the training pipeline 1404. In at least one embodiment, system 1400 may include a multi-tier platform that may include a diagnostic application (or other application type) software layer (e.g., software 1318) that may perform one or more medical imaging and diagnostic functions. In at least one embodiment, system 1400 may be communicatively coupled (e.g., via an encrypted link) to one or more institutional PACS server networks.In at least one embodiment, system 1400 may be configured to access and reference data from a PACS server to perform operations such as training machine learning models, deploying machine learning models, image processing, inference, and / or other operations.

[0147] In at least one embodiment, the software layer may be implemented as a secure, encrypted, and / or authenticated API through which applications or containers may be invoked (e.g., called) from external environment(s) (e.g., facility 1302). In at least one embodiment, the applications may then call or execute one or more services 1320 to perform compute, AI, or visualization tasks associated with the respective application, and software 1318 and / or services 1320 may leverage hardware 1322 to perform the processing tasks in an effective and efficient manner.

[0148] In at least one embodiment, the deployment system 1306 may execute an deployment pipeline 1410. In at least one embodiment, the deployment pipeline 1410 may include any number of applications, which may be applied sequentially, non-sequentially, or otherwise to imaging data (and / or other data types) generated by an imaging device, a sequencing device, a genomics device, etc., including AI-assisted annotation, as described above. In at least one embodiment, as described herein, the deployment pipeline 1410 for an individual device may be referred to as a virtual instrument for the device (e.g., a virtual ultrasound instrument, a virtual CT scan instrument, a virtual sequencing instrument, etc.). In at least one embodiment, there may be two or more deployment pipelines 1410 for a single device, depending on the information desired from the data generated by the device. In at least one embodiment, there may be a first deployment pipeline 1410 if anomaly detection is desired from an MRI machine, and a second deployment pipeline 1410 if image enhancement is desired from the output of the MRI machine.

[0149] In at least one embodiment, the image generation application may include processing tasks that involve the use of machine learning models. In at least one embodiment, a user may desire to use their own machine learning model or select a machine learning model from model registry 1324. In at least one embodiment, a user may implement their own machine learning model or select a machine learning model for inclusion in an application to perform a processing task. In at least one embodiment, applications may be selectable and customizable, and by defining the architecture of an application, the deployment and implementation of the application for a particular user may be presented as a more seamless user experience. In at least one embodiment, by leveraging other features of system 1400, such as services 1320 and hardware 1322, deployment pipeline 1410 may be even more user-friendly, provide easier integration, and produce more accurate, efficient, and timely results.

[0150] In at least one embodiment, the deployment system 1306 may include a user interface 1414 (e.g., a graphical user interface, a web interface, etc.) that may be used to select applications for inclusion in the deployment pipeline(s) 1410, deploy applications, modify or change applications or their parameters or configuration, use and interact with the deployment pipeline(s) 1410 during setup and / or deployment, and / or otherwise interact with the deployment system 1306. In at least one embodiment, although not shown with respect to the training system 1304, the user interface 1414 (or a different user interface) may be used to select models for use in the deployment system 1306, to select models for training or retraining in the training system 1304, and / or to otherwise interact with the training system 1304.

[0151] In at least one embodiment, a pipeline manager 1412 may be used in addition to an application orchestration system 1428 to manage interactions between applications or containers of deployment pipeline(s) 1410 and services 1320 and / or hardware 1322. In at least one embodiment, the pipeline manager 1412 may be configured to facilitate application-to-application interactions, application-to-service interactions, and / or application or service-to-hardware interactions. While shown in at least one embodiment as being included in software 1318, this is not intended to be limiting, and in some instances (e.g., shown in FIG. 12C), the pipeline manager 1412 may be included in services 1320. In at least one embodiment, the application orchestration system 1428 (e.g., Kubernetes, DOCKER, etc.) may include a container orchestration system that may group applications into containers as logical units for coordination, management, scaling, and deployment. In at least one embodiment, by associating applications (e.g., reconstruction applications, segmentation applications, etc.) from the deployment pipeline(s) 1410 with individual containers, each application may run in a self-contained environment (e.g., at the kernel level) for improved speed and efficiency.

[0152] In at least one embodiment, each application and / or container (or image thereof) may be developed, modified, and deployed individually (e.g., a first user or developer may develop, modify, and deploy a first application, and a second user or developer may develop, modify, and deploy a second application separately from the first user or developer), which may allow for focused attention to be paid to the tasks of a single application and / or container(s) without being distracted by the tasks of another application(s) or container(s). In at least one embodiment, communication and coordination between different containers or applications may be assisted by pipeline manager 1412 and application orchestration system 1428. In at least one embodiment, application orchestration system 1428 and / or pipeline manager 1412 may facilitate communication between, and sharing of resources between, each of the applications or containers, as long as the expected inputs and / or outputs of each container or application are known by the system (e.g., based on the construction of the application or container). In at least one embodiment, one or more of the applications or containers in the deployment pipeline(s) 1410 may share the same services and resources, and therefore, the application orchestration system 1428 may orchestrate, load balance, and determine sharing of services or resources among the various applications or containers. In at least one embodiment, a scheduler may be used to track the resource requirements of the applications or containers, the current or planned usage of those resources, and the availability of the resources. Thus, in at least one embodiment, the scheduler may allocate resources to different applications and distribute the resources among the applications taking into account the requirements and availability of the system.In some instances, the scheduler (and / or other components of the application orchestration system 1428) may determine resource availability and distribution based on constraints imposed on the system (e.g., user constraints), such as quality of service (QoS), the urgency of needing data output (e.g., to determine whether real-time or delayed processing should be performed), etc.

[0153] In at least one embodiment, services 1320 utilized and shared by applications or containers in deployment system 1306 may include compute services 1416, AI services 1418, visualization services 1420, and / or other service types. In at least one embodiment, an application may call (e.g., execute) one or more of services 1320 to perform processing operations for the application. In at least one embodiment, compute services 1416 may be utilized by an application to perform supercomputing or other high-performance computing (HPC) tasks. In at least one embodiment, compute service(s) 1416 may be utilized to perform parallel processing (e.g., using parallel computing platform 1430) to process data through one or more of the applications and / or one or more tasks of a single application substantially simultaneously. In at least one embodiment, a parallel computing platform 1430 (e.g., NVIDIA's CUDA) may enable general purpose computing on GPUs (e.g., GPU 1422). In at least one embodiment, a software layer of the parallel computing platform 1430 may provide a virtual instruction set and access to the parallel computational elements of the GPU for execution of compute kernels. In at least one embodiment, the parallel computing platform 1430 may include memory, which in some embodiments may be shared among multiple containers and / or among different processing tasks within a single container.In at least one embodiment, inter-process communication (IPC) calls may be generated for multiple containers and / or multiple processes within a container to use the same data from a shared segment of memory in the parallel computing platform 1430 (e.g., when multiple different stages of an application or multiple applications are processing the same information). In at least one embodiment, rather than making copies of the data and moving the data to different locations in memory (e.g., read / write operations), the same data in the same location in memory may be used for any number of processing tasks (e.g., at the same time, different times, etc.). In at least one embodiment, as data is used and new data is generated as a result of processing, this information of the data's new location may be stored and shared among various applications. In at least one embodiment, the location of the data and the location of updated or modified data may be part of the definition of how the payload is understood within the container.

[0154] In at least one embodiment, AI service 1418 may be utilized to perform inference services to execute machine learning model(s) associated with an application (e.g., tasked with performing one or more processing tasks of the application). In at least one embodiment, AI service 1418 may utilize AI system 1424 to execute machine learning model(s) (e.g., neural networks such as CNNs) for segmentation, reconstruction, object detection, feature detection, classification, and / or other inference tasks. In at least one embodiment, applications in deployment pipeline(s) 1410 may perform inference on imaging data using output model 1316 from training system 1304 and / or one or more of the application's other models. In at least one embodiment, two or more instances of inference using application orchestration system 1428 (e.g., a scheduler) may be available. In at least one embodiment, the first category may include a high priority / low latency path that may achieve a higher service level agreement, such as for performing inference on urgent requests during an emergency or for radiologists during a diagnosis. In at least one embodiment, the second category may include a standard priority path that may be used for requests that may not be urgent or where analysis can be performed at a later time. In at least one embodiment, the application orchestration system 1428 may distribute resources (e.g., services 1320 and / or hardware 1322) based on the priority path for different inference tasks of the AI ​​service 1418.

[0155] In at least one embodiment, shared storage may be attached to the AI ​​service 1418 within the system 1400. In at least one embodiment, the shared storage may act as a cache (or other storage device type) and may be used to process inference requests from applications. In at least one embodiment, when an inference request is submitted, the request may be received by a set of API instances of the installation system 1306, and one or more instances may be selected (e.g., for best fit, for load balancing, etc.) to process the request. In at least one embodiment, to process the request, the request may be entered into a database, the machine learning model may be located from the model registry 1324 if not already in the cache, and a validation step may ensure that the appropriate machine learning model is loaded into the cache (e.g., shared storage) and / or a copy of the model may be saved to the cache. In at least one embodiment, if the application is not already running or if there are not enough instances of the application, a scheduler (e.g., of the pipeline manager 1412) may be used to launch the application referenced in the request. In at least one embodiment, an inference server for executing a model may be started if it has not already been started. Any number of inference servers may be started per model. In at least one embodiment, in a pull model where inference servers are clustered, models may be cached whenever load balancing is advantageous. In at least one embodiment, inference servers may be statically loaded onto corresponding distributed servers.

[0156] In at least one embodiment, inference may be performed using an inference server running in a container. In at least one embodiment, an instance of an inference server may be associated with a model (optionally, multiple versions of the model). In at least one embodiment, when a request to perform inference on a model is received, if an instance of the inference server does not exist, a new instance may be loaded. In at least one embodiment, when the inference server is started, the model may be passed to the inference server, such that the same container may be used to serve different models as long as the inference servers are running as different instances.

[0157] In at least one embodiment, during application execution, an inference request for a given application may be received, a container (e.g., hosting an instance of an inference server) may be loaded (if not already loaded), and a start procedure may be called. In at least one embodiment, preprocessing logic in the container may load, decode, and / or perform any additional preprocessing on the incoming data (e.g., using CPU(s) and / or GPU(s)). In at least one embodiment, once the data is prepared for inference, the container may perform inference on the data as needed. In at least one embodiment, this may involve a single inference call for one image (e.g., a hand X-ray) or may require inference on hundreds of images (e.g., a chest CT). In at least one embodiment, the application may summarize results before completion, which may include, but is not limited to, generating a single confidence score, pixel-level segmentation, voxel-level segmentation, visualization, or generating text to summarize findings. In at least one embodiment, different models or applications may be assigned different priorities. For example, some models may have real-time (TAT<1 minute) priority, while other models may have low priority (e.g., TAT<10 minutes). In at least one embodiment, model execution time may be measured from the requesting institution or entity and may include partner network traversal time, as well as execution against the inference service.

[0158] In at least one embodiment, the transfer of requests between the service 1320 and the inference application may be hidden behind a software development kit (SDK), and robust transport may be provided through queues. In at least one embodiment, requests are queued via an API for each application / tenant ID combination, and the SDK pulls the requests from the queue and provides them to the application. In at least one embodiment, the name of the queue may be provided in the environment where the SDK picks up the request. In at least one embodiment, asynchronous communication through queues may be useful because it may allow any instance of the application to pick up work when it becomes available. Results may be sent back through queues to ensure that data is not lost. In at least one embodiment, queues may also provide the ability to segment work, as the highest priority work may go to a queue with most instances of the application attached to it, while the lowest priority work may go to a queue with a single instance attached to it, processing tasks in the order they were received. In at least one embodiment, the application may run on a GPU-accelerated instance created in the cloud 1426, and the inference service may perform inference on the GPU.

[0159] In at least one embodiment, a visualization service 1420 may be utilized to generate visualizations for viewing the output of the application and / or the deployment pipeline(s) 1410. In at least one embodiment, a GPU 1422 may be utilized by the visualization service 1420 to generate the visualization. In at least one embodiment, rendering effects such as ray tracing may be implemented by the visualization service 1420 to generate higher quality visualizations. In at least one embodiment, the visualization may include, but is not limited to, 2D image rendering, 3D volume rendering, 3D volume reconstruction, 2D tomography slices, virtual reality displays, augmented reality displays, etc. In at least one embodiment, a virtualized environment may be used to generate a virtual interactive display or environment (e.g., a virtual environment) for interaction by a user of the system (e.g., a doctor, nurse, radiologist, etc.). In at least one embodiment, visualization services 1420 may include internal visualizer, cinematics, and / or other rendering or image processing capabilities or functionality (e.g., ray tracing, rasterization, internal optics, etc.).

[0160] In at least one embodiment, hardware 1322 may include GPU 1422, AI system 1424, cloud 1426, and / or any other hardware used to run training system 1304 and / or deployment system 1306. In at least one embodiment, GPU 1422 (e.g., NVIDIA TESLA and / or QUADRO GPUs) may include any number of GPUs that may be used to perform processing tasks of compute services 1416, AI services 1418, visualization services 1420, other services, and / or any of the features or functionality of software 1318. For example, with respect to AI services 1418, GPU 1422 may be used to perform pre-processing on imaging data (or other data types used by machine learning models), post-processing on the output of machine learning models, and / or to perform inference (e.g., to run machine learning models). In at least one embodiment, cloud 1426, AI system 1424, and / or other components of system 1400 may use GPUs 1422. In at least one embodiment, cloud 1426 may include a GPU-optimized platform for deep learning tasks. In at least one embodiment, AI system 1424 may use GPUs, and cloud 1426, or at least a portion tasked with deep learning or inference, may be executed using one or more AI systems 1424. Thus, while hardware 1322 is shown as discrete components, this is not intended to be limiting, and any component of hardware 1322 may be combined with or utilized by any other component of hardware 1322.

[0161] In at least one embodiment, AI system 1424 may include a specialized computing system (e.g., a supercomputer or HPC) configured for inference, deep learning, machine learning, and / or other artificial intelligence tasks. In at least one embodiment, AI system 1424 (e.g., NVIDIA's DGX) may include GPU optimization software (e.g., a software stack) that may be executed using multiple GPUs 1422 in addition to CPUs, RAM, storage, and / or other components, features, or functionality. In at least one embodiment, one or more AI systems 1424 may be implemented in cloud 1426 (e.g., in a data center) to perform some or all of the AI-based processing tasks of system 1400.

[0162] In at least one embodiment, cloud 1426 may include a GPU-accelerated infrastructure (e.g., NVIDIA's NGC), which may provide a GPU-optimized platform for executing processing tasks of system 1400. In at least one embodiment, cloud 1426 may include AI system(s) 1424 (e.g., as a hardware abstraction and scaling platform) for performing one or more of the AI-based tasks of system 1400. In at least one embodiment, cloud 1426 may leverage multiple GPUs and integrate with application orchestration system 1428 to enable seamless scaling and load balancing between applications and services 1320. In at least one embodiment, cloud 1426 may be tasked with running at least some of the services 1320 of system 1400, including compute services 1416, AI services 1418, and / or visualization services 1420, as described herein. In at least one embodiment, cloud 1426 may perform large or small batch inference (e.g., running NVIDIA's TENSOR RT), provide an accelerated parallel computing API and platform 1430 (e.g., NVIDIA's CUDA), run an application orchestration system 1428 (e.g., KUBERNETES), provide a graphics rendering API and platform (e.g., for ray tracing, 2D graphics, 3D graphics, and / or other rendering techniques to produce higher quality cinematics), and / or provide other functionality for system 1400.

[0163] 15A shows a data flow diagram for a process 1500 for training, retraining, or updating a machine learning model, according to at least one embodiment. In at least one embodiment, process 1500 may be performed using system 1400 of FIG. 14 as a non-limiting example. In at least one embodiment, process 1500 may leverage services 1320 and / or hardware 1322 of system 1400, as described herein. In at least one embodiment, improved model 1512 generated by process 1500 may be executed by deployment system 1306 for one or more containerized applications in deployment pipeline 1410.

[0164] In at least one embodiment, model training 1314 may include retraining or updating the initial model 1504 (e.g., a pre-trained model) using new training data (e.g., new input data, such as the customer dataset 1506 and / or new ground truth data associated with the input data). In at least one embodiment, to retrain or update the initial model 1504, output or loss layer(s) of the initial model 1504 may be reset or removed and / or replaced with updated or new output or loss layer(s). In at least one embodiment, the initial model 1504 may have previously fine-tuned parameters (e.g., weights and / or biases) remaining from previous training; therefore, training or retraining 1314 may not take as long or require as much processing as training a model from scratch. In at least one embodiment, by having the output or loss layer(s) of the initial model 1504 reset or replaced during model training 1314, parameters can be updated or retuned for the new data set based on a loss calculation associated with the accuracy of the output or loss layer(s) in generating predictions for the new customer data set 1506 (e.g., image data 1308 in FIG. 13 ).

[0165] In at least one embodiment, the pre-trained model 1406 may be stored in a data store or registry (e.g., model registry 1324 of FIG. 13 ). In at least one embodiment, the pre-trained model 1406 may have been trained, at least in part, at one or more facilities other than the facility performing process 1500. In at least one embodiment, to protect the privacy and rights of patients, subjects, or customers of different facilities, the pre-trained model 1406 may have been trained on-premise using customer or patient data generated on-premise. In at least one embodiment, the pre-trained model 1406 may be trained using the cloud 1426 and / or other hardware 1322, but privacy-protected sensitive patient data may not be transferred to, used by, or accessible to any components of the cloud 1426 (or other off-premise hardware). In at least one embodiment, if pre-trained model 1406 is trained using patient data from more than one facility, pre-trained model 1406 may be trained individually for each facility and then trained on patient or customer data from another facility. In at least one embodiment, customer or patient data from any number of facilities may be used to train pre-trained model 1406 on-premise and / or off-premise, such as in a data center or other cloud computing infrastructure, if the customer or patient data is free from privacy concerns (e.g., via a waiver, for experimental use, etc.) or if the customer or patient data is included in a public data set.

[0166] In at least one embodiment, when selecting an application for use in the deployment pipeline 1410, a user may also select a machine learning model to be used for the particular application. In at least one embodiment, a user may not have a model for use, and therefore the user may select a pre-trained model 1406 for use with the application. In at least one embodiment, the pre-trained model 1406 may not be optimized to produce accurate results for the user's facility's customer dataset 1506 (e.g., based on patient diversity, demographics, type of medical imaging device used, etc.). In at least one embodiment, before deploying the pre-trained model 1406 in the deployment pipeline 1410 for use with application(s), the pre-trained model 1406 may be updated, retrained, and / or fine-tuned for use at the respective facility.

[0167] In at least one embodiment, a user may select a pre-trained model 1406 to be updated, retrained, and / or fine-tuned; the pre-trained model 1406 may be referred to as an initial model 1504 for the training system 1304 in process 1500. In at least one embodiment, a customer dataset 1506 (e.g., imaging data, genomics data, sequencing data, or other data types generated by devices at the facility) may be used to perform model training 1314 (which may include, but is not limited to, transfer learning) on ​​the initial model 1504 to generate an improved model 1512. In at least one embodiment, ground truth data corresponding to the customer dataset 1506 may be generated by the training system 1304. In at least one embodiment, the ground truth data may be generated, at least in part, by clinicians, scientists, physicians, or practitioners at the facility (e.g., as labeled clinic data 1312 of FIG. 13 ).

[0168] In at least one embodiment, AI-assisted annotation 1310 may be used in some instances to generate ground truth data. In at least one embodiment, AI-assisted annotation 1310 (e.g., implemented using an AI-assisted annotation SDK) may utilize machine learning models (e.g., neural networks) to generate suggested or predicted ground truth data for a customer dataset. In at least one embodiment, user 1510 may use annotation tools within a user interface (graphical user interface (GUI)) on computing device 1508.

[0169] In at least one embodiment, a user 1510 may interact with a GUI via a computing device 1508 to edit or fine-tune the (automatic) annotation. In at least one embodiment, a polygon editing feature may be used to move vertices of a polygon to more precise or fine-tuned locations.

[0170] In at least one embodiment, once the customer dataset 1506 has associated ground truth data, the ground truth data (e.g., from AI-assisted annotation, manual labeling, etc.) may be used during model training 1314 to generate an improved model 1512. In at least one embodiment, the customer dataset 1506 may be applied to the initial model 1504 any number of times, and the ground truth data may be used to update parameters of the initial model 1504 until an acceptable level of accuracy is achieved for the improved model 1512. In at least one embodiment, once the improved model 1512 is generated, the improved model 1512 may be deployed in one or more deployment pipelines 1410 at a facility to perform one or more processing tasks on the medical imaging data.

[0171] In at least one embodiment, the improved model 1512 may be uploaded to the pre-trained models 1406 in the model registry 1324 to be selected by another facility. In at least one embodiment, this process may be completed at any number of facilities, such that the improved model 1512 may be further improved any number of times on new datasets to produce a more generalized model.

[0172] FIG. 15B is an example diagram of a client-server architecture 1532 for extending an annotation tool with a pre-trained annotation model, according to at least one embodiment. In at least one embodiment, an AI-assisted annotation tool 1536 may be instantiated based on the client-server architecture 1532. In at least one embodiment, the annotation tool 1536 in an imaging application may assist a radiologist, for example, in identifying organs and abnormalities. In at least one embodiment, the imaging application may include, by way of non-limiting example, a software tool that helps a user 1510 identify a few extreme points on a particular organ of interest in a raw image 1534 (e.g., in a 3D MRI or CT scan) and receive automatically annotated results for all 2D slices of the particular organ. In at least one embodiment, the results are stored in a data store as training data 1538 and may be used (for example, without limitation) as ground truth data for training. In at least one embodiment, when computing device 1508 sends extremum points for AI-assisted annotation 1310, for example, a deep learning model may receive this data as input and return inferences of segmented organs or anomalies. In at least one embodiment, a pre-instantiated annotation tool, such as AI-assisted annotation tool 1536B in FIG. 15B , may be extended by making API calls (e.g., API call 1544) to a server, such as annotation-assisted server 1540, which may include a set of pre-trained models 1542 stored in an annotation model registry. In at least one embodiment, annotation model registry may store pre-trained models 1542 (e.g., machine learning models such as deep learning models) that have been pre-trained to perform AI-assisted annotation for specific organs or anomalies. These models may be further updated using training pipeline 1404.In at least one embodiment, the pre-installed annotation tools may be improved over time as new labeled clinic data 1312 is added.

[0173] Such components may be used to combine the constituent images into a single representation using parameters determined from one or more quality assessment values.

[0174] Automated Technology FIG. 16A is a block diagram illustrating an example system architecture for the autonomous vehicle 1600 of FIG. 16A , according to at least one embodiment. In at least one embodiment, each of the components, features, and systems of the vehicle 1600 in FIG. 16A is shown as connected via a bus 1602. In at least one embodiment, the bus 1602 may include, but is not limited to, a CAN data interface (alternatively referred to herein as a “CAN bus”). In at least one embodiment, the CAN bus may be a network internal to the vehicle 1600 used to help control various features and functionality of the vehicle 1600, such as brake application, acceleration, brake control, steering, windshield wipers, etc. In at least one embodiment, the bus 1602 may be configured to have tens or even hundreds of nodes, each with its own unique identifier (e.g., a CAN ID). In at least one embodiment, bus 1602 can be read to determine steering angle, ground speed, engine revolutions per minute ("RPM"), button position, and / or other vehicle status indicators. In at least one embodiment, bus 1602 can be an ASIL B compliant CAN bus.

[0175] In at least one embodiment, FlexRay and / or Ethernet may be used in addition to or as an alternative to CAN. In at least one embodiment, there may be any number of buses 1602, including, but not limited to, zero or more CAN buses, zero or more FlexRay buses, zero or more Ethernet buses, and / or zero or more other types of buses using different protocols. In at least one embodiment, two or more buses 1602 may be used to perform different functions and / or for redundancy. For example, a first bus 1602 may be used for collision avoidance functionality, and a second bus 1602 may be used for actuation control. In at least one embodiment, each bus 1602 may communicate with any of the components of the vehicle 1600, and two or more buses 1602 may communicate with the same component. In at least one embodiment, each of any number of system-on-chip (“SoC”) 1604, each of the controller(s) 1636, and / or each computer in the vehicle may have access to the same input data (e.g., input from sensors in the vehicle 1600) and may be connected to a common bus, such as a CAN bus.

[0176] In at least one embodiment, vehicle 1600 may include one or more controllers 1636, such as those described herein with respect to FIG. 1A. Controller(s) 1636 may be used for a variety of functions. In at least one embodiment, controller(s) 1636 may be coupled to any of various other components and systems of vehicle 1600 and may be used for control of vehicle 1600, artificial intelligence of vehicle 1600, infotainment for vehicle 1600, etc.

[0177] In at least one embodiment, vehicle 1600 may include any number of SoCs 1604. Each of SoCs 1604 may include, without limitation, a central processing unit ("CPU") 1606, a graphics processing unit ("GPU") 1608, processor(s) 1610, cache(s) 1612, accelerator(s) 1614, data store(s) 1616, and / or other components and features not shown. In at least one embodiment, SoC(s) 1604 may be used to control vehicle 1600 in a variety of platforms and systems. For example, in at least one embodiment, the SoC(s) 1604 may be combined in a system (e.g., a system in a vehicle 1600) with a high definition (“HD”) map 1622 that may obtain map refreshes and / or updates via a network interface 1624 from one or more servers (not shown in FIG. 16A).

[0178] In at least one embodiment, the CPU(s) 1606 may include a CPU cluster or CPU complex (alternatively referred to herein as a “CCPLEX”). In at least one embodiment, the CPU(s) 1606 may include multiple cores and / or level 2 (“L2”) caches. For example, in at least one embodiment, the CPU(s) 1606 may include eight cores in a coherent multiprocessor configuration. In at least one embodiment, the CPU(s) 1606 may include four dual-core clusters, each cluster having a dedicated L2 cache (e.g., 2 MB of L2 cache). In at least one embodiment, the CPU(s) 1606 (e.g., a CCPLEX) may be configured to support simultaneous cluster operation, allowing any combination of clusters of the CPU(s) 1606 to be active at any given time.

[0179] In at least one embodiment, one or more of the CPU(s) 1606 may implement power management capabilities including, but not limited to, one or more of the following features: individual hardware blocks may be automatically clock gated when idle to conserve dynamic power; each core clock may be gated when the core is not actively executing instructions by executing a Wait for Interrupt ("WFI") / Wait for Event ("WFE") instruction; each core may be independently power gated; each core cluster may be independently clock gated when all cores are clock gated or power gated; and / or each core cluster may be independently power gated when all cores are power gated. In at least one embodiment, the CPU(s) 1606 may further implement an advanced algorithm for managing power states, where allowed power states and expected wake-up times are specified and the hardware / microcode determines the best power state to enter for the core, cluster, and CCPLEX. In at least one embodiment, the processing core may support a simple power state entry sequence in software with work offloaded to microcode.

[0180] In at least one embodiment, the GPU(s) 1608 may include an integrated GPU (alternatively referred to herein as an “iGPU”). In at least one embodiment, the GPU(s) 1608 may be programmable and efficient for parallel workloads. In at least one embodiment, the GPU(s) 1608 may use an extended tensor instruction set. In at least one embodiment, the GPU(s) 1608 may include one or more streaming microprocessors, each of which may include a level 1 (“L1”) cache (e.g., an L1 cache with at least 96 KB of storage capacity), and two or more of the streaming microprocessors may share an L2 cache (e.g., an L2 cache with 512 KB of storage capacity). In at least one embodiment, the GPU(s) 1608 may include at least eight streaming microprocessors. In at least one embodiment, the GPU(s) 1608 may use one or more compute application programming interfaces (APIs). In at least one embodiment, the GPU(s) 1608 may use one or more parallel computing platforms and / or programming models (e.g., NVIDIA's CUDA).

[0181] In at least one embodiment, one or more of the GPU(s) 1608 may be power-optimized for best performance in automotive and embedded use cases. For example, in one embodiment, the GPU(s) 1608 may be fabricated on Fin field-effect transistors ("FinFETs"). In at least one embodiment, each streaming microprocessor may incorporate several mixed-precision processing cores partitioned into multiple blocks. For example, without limitation, 64 PF32 cores and 32 PF64 cores may be partitioned into four processing blocks. In at least one embodiment, each processing block may be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed-precision NVIDIA TENSOR COREs for deep learning matrix arithmetic, a level 0 ("L0") instruction cache, a warp scheduler, a dispatch unit, and / or a 64KB register file. In at least one embodiment, the streaming microprocessor may include independent parallel integer and floating-point data paths for efficient execution of workloads with a mix of computation and addressing calculations. In at least one embodiment, the streaming microprocessor may include independent thread scheduling capabilities to enable finer-grained synchronization and cooperation among parallel threads. In at least one embodiment, the streaming microprocessor may include a combined L1 data cache and shared memory unit to improve performance while simplifying programming.

[0182] In at least one embodiment, one or more of the GPU(s) 1608 may include high bandwidth memory (“HBM”) and / or a 16 GB HBM2 memory subsystem to provide, in some instances, approximately 900 GB / s of peak memory bandwidth. In at least one embodiment, synchronous graphics random-access memory (“SGRAM”), such as graphics double data rate type five synchronous random-access memory (“GDDR5”), may be used in addition to or as an alternative to the HBM memory.

[0183] In at least one embodiment, the GPU(s) 1608 may include unified memory technology. In at least one embodiment, address translation service ("ATS") support may be used to allow the GPU(s) 1608 to directly access the page tables of the CPU(s) 1606. In at least one embodiment, when the memory management unit ("MMU") of the GPU(s) 1608 encounters a miss, an address translation request may be sent to the CPU(s) 1606. In at least one embodiment, in response, the CPU(s) 1606 may look up a virtual-to-physical mapping for the address in its page table and send the translation back to the GPU(s) 1608. In at least one embodiment, the unified memory technology enables a single unified virtual address space for memory of both the CPU(s) 1606 and the GPU(s) 1608, which may simplify programming the GPU(s) 1608 and porting applications to the GPU(s) 1608.

[0184] In at least one embodiment, the GPU(s) 1608 may include any number of access counters that may track the frequency of the GPU(s) 1608's accesses to the memory of other processors. In at least one embodiment, the access counter(s) may help ensure that memory pages are moved to the physical memory of the processor that is accessing the pages most frequently, thereby improving the efficiency of memory ranges shared between processors.

[0185] In at least one embodiment, one or more of the SoC(s) 1604 may include any number of caches 1612, including those described herein. For example, in at least one embodiment, the cache(s) 1612 may include a level 3 (“L3”) cache that is available to both the CPU(s) 1606 and the GPU(s) 1608 (e.g., connected to both the CPU(s) 1606 and the GPU(s) 1608). In at least one embodiment, the cache(s) 1612 may include a write-back cache that may track line states, such as by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). In at least one embodiment, the L3 cache may include 4 MB or more, depending on the embodiment, although smaller cache sizes may be used.

[0186] In at least one embodiment, one or more of the SoC(s) 1604 may include one or more accelerators 1614 (e.g., hardware accelerators, software accelerators, or a combination thereof). In at least one embodiment, the SoC(s) 1604 may include a hardware acceleration cluster, which may include optimized hardware accelerators and / or large on-chip memory. In at least one embodiment, large on-chip memory (e.g., 4 MB of SRAM) may enable the hardware acceleration cluster to accelerate neural networks and other calculations. In at least one embodiment, the hardware acceleration cluster may complement the GPU(s) 1608 and be used to offload some of the GPU(s) 1608's tasks (e.g., to free up more cycles of the GPU(s) 1608 to perform other tasks). In at least one embodiment, accelerator(s) 1614 may be used for workloads of interest that are stable enough to accommodate acceleration (e.g., perception, convolutional neural networks (“CNNs”), recurrent neural networks (“RNNs”), etc.). In at least one embodiment, the CNNs may include region-based, i.e., regional convolutional neural networks (“RCNNs”), and Fast RCNNs (e.g., as used for object detection), or other types of CNNs.

[0187] In at least one embodiment, the accelerator(s) 1614 (e.g., a hardware acceleration cluster) may include deep learning accelerator(s) (“DLA”). The DLA(s) may include, without limitation, one or more tensor processing units (“TPU”), which may be configured to provide an additional tens of trillion operations per second for deep learning applications and inference. In at least one embodiment, the TPU(s) may be accelerators configured and optimized for performing image processing functions (e.g., for CNN, RCNN, etc.). The DLA(s) may be further optimized for a specific set of neural network types and floating-point operations, as well as for inference. In at least one embodiment, the design of the DLA(s) may provide more performance per millimeter than a typical general-purpose GPU, generally far exceeding the performance of a CPU. In at least one embodiment, the TPU(s) may perform several functions, including, for example, single-instance convolution functions, supporting INT8, INT16, and FP16 data types for both features and weights, and post-processor functions. In at least one embodiment, the DLA(s) may quickly and efficiently run neural networks, particularly CNNs, on processed or unprocessed data for any of a variety of functions, including, for example, but not limited to, CNNs for object identification and detection using data from a camera sensor, CNNs for distance estimation using data from a camera sensor, CNNs for emergency vehicle detection and identification using data from microphone 1696, CNNs for face recognition and vehicle owner identification using data from a camera sensor, and / or CNNs for security and / or safety-related events.

[0188] In at least one embodiment, the DLA(s) may perform any function of the GPU(s) 1608; for example, by using an inference accelerator, a designer may target either the DLA(s) or the GPU(s) 1608 for any function. For example, in at least one embodiment, a designer may centralize the processing of CNNs and floating-point operations in the DLA(s) and offload other functions to the GPU(s) 1608 and / or other accelerator(s) 1614.

[0189] In at least one embodiment, the accelerator(s) 1614 (e.g., a hardware acceleration cluster) may include programmable vision accelerator(s), which may alternatively be referred to herein as computer vision accelerators. In at least one embodiment, the PVA(s) may be designed and configured to accelerate computer vision algorithms for advanced driver assistance systems ("ADAS") 1638, autonomous driving, augmented reality ("AR") applications, and / or virtual reality ("VR") applications. The PVA(s) may provide a balance between performance and flexibility. For example, in at least one embodiment, each PVA(s) may include, for example, without limitation, any number of reduced instruction set computer ("RISC") cores, direct memory access ("DMA") processors, and / or any number of vector processors.

[0190] In at least one embodiment, the RISC core may interact with an image sensor (e.g., an image sensor of any of the cameras described herein), image signal processor(s), etc. In at least one embodiment, each of the RISC cores may include any amount of memory. In at least one embodiment, the RISC cores may use any of several protocols, depending on the embodiment. In at least one embodiment, the RISC cores may execute a real-time operating system (“RTOS”). In at least one embodiment, the RISC cores may be implemented using one or more integrated circuit devices, application specific integrated circuits (“ASICs”), and / or memory devices. For example, in at least one embodiment, the RISC cores may include an instruction cache and / or tightly coupled RAM.

[0191] In at least one embodiment, the DMA may enable components of the PVA(s) to access system memory independently of the CPU(s) 1606. In at least one embodiment, the DMA may support any number of features used to provide optimizations to the PVA, including, but not limited to, supporting multi-dimensional addressing and / or circular addressing. In at least one embodiment, the DMA may support up to six or more addressing dimensions, which may include, but are not limited to, block width, block height, block depth, horizontal block stepping, vertical block stepping, and / or depth stepping.

[0192] In at least one embodiment, the vector processor is a programmable processor that can be designed to efficiently and flexibly execute programming for computer vision algorithms and can provide signal processing capabilities. In at least one embodiment, the PVA can include a PVA core and two vector processing subsystem partitions. In at least one embodiment, the PVA core can include a processor subsystem, one or more DMA engines (e.g., two DMA engines), and / or other peripherals. In at least one embodiment, the vector processing subsystem can act as the PVA's primary processing engine and can include a vector processing unit ("VPU"), an instruction cache, and / or vector memory (e.g., "VMEM"). In at least one embodiment, the VPU can include a digital signal processor, such as a single instruction, multiple data ("SIMD"), very long instruction word ("VLIW") digital signal processor. In at least one embodiment, the combination of SIMD and VLIW can improve throughput and speed.

[0193] In at least one embodiment, each of the vector processors may include an instruction cache and may be coupled to dedicated memory. As a result, in at least one embodiment, each of the vector processors may be configured to execute independently of the other vector processors. In at least one embodiment, the vector processors included in a particular PVA may be configured to employ data parallelism. For example, in at least one embodiment, multiple vector processors included in a single PVA may execute the same computer vision algorithm, but on different regions of an image. In at least one embodiment, the vector processors included in a particular PVA may execute different computer vision algorithms simultaneously on the same image, or even execute different algorithms on consecutive images or portions of an image. In at least one embodiment, among other things, any number of PVAs may be included in a hardware-accelerated cluster, and any number of vector processors may be included within each PVA. In at least one embodiment, the PVA(s) may include additional error correction code ("ECC") memory to improve the overall security of the system.

[0194] In at least one embodiment, the accelerator(s) 1614 (e.g., a hardware acceleration cluster) may include a computer vision network-on-chip and static random access memory (“SRAM”) to provide high-bandwidth, low-latency SRAM for the accelerator(s) 1614. In at least one embodiment, the on-chip memory may include, for example, but not limited to, at least 4 MB of SRAM consisting of eight field-configurable memory blocks, which may be accessible by both the PVA and DLA. In at least one embodiment, each pair of memory blocks may include an advanced peripheral bus (“APB”) interface, configuration circuitry, a controller, and a multiplexer. In at least one embodiment, any type of memory may be used. In at least one embodiment, the PVA and DLA may access memory through a backbone that provides the PVA and DLA with high-speed access to memory. In at least one embodiment, the backbone may include a computer vision network-on-chip that interconnects the PVA and DLA to memory (e.g., using APBs).

[0195] In at least one embodiment, the computer vision network-on-chip may include an interface that determines whether both the PVA and DLA provide ready and enable signals before transmitting any control signals / addresses / data. In at least one embodiment, the interface may provide separate phases and separate channels for transmitting control signals / addresses / data, as well as burst-type communication for continuous data transfer. In at least one embodiment, the interface may conform to International Organization for Standardization (“ISO”) 26262 or International Electrotechnical Commission (“IEC”) 61508 standards, although other standards and protocols may be used.

[0196] In at least one embodiment, one or more of the SoC1604(s) may include a real-time ray tracing hardware accelerator that may be used to quickly and efficiently determine the location and range of objects (e.g., within a world model) to generate real-time visualization simulations for RADAR signal interpretation, for sound propagation synthesis and / or analysis, for SONAR system simulation, for general waveform propagation simulation, for comparison with LIDAR data for localization and / or other functions, and / or other uses.

[0197] In at least one embodiment, accelerator(s) 1614 (e.g., a hardware accelerator cluster) have diverse uses for autonomous driving. In at least one embodiment, the PVA can be a programmable vision accelerator that can be used for key processing stages in ADAS and autonomous vehicles. In at least one embodiment, the capabilities of the PVA are well-matched for algorithm domains that require predictable processing with low power and low latency. In other words, the PVA performs well on small data sets for semi-dense or dense regular calculations that require predictable runtime with low latency and low power. In at least one embodiment, an autonomous vehicle, such as vehicle 1600, the PVA is designed to run traditional computer vision algorithms because they are efficient at object detection and integer arithmetic.

[0198] For example, according to at least one embodiment of the technology, the PVA is used to perform computer stereo vision. In at least one embodiment, semi-global matching-based algorithms may be used in some instances, but this is not intended to be limiting. In at least one embodiment, applications for Level 3-5 autonomous driving use motion estimation / stereo matching (e.g., structure from motion, pedestrian recognition, lane detection, etc.) on the fly. In at least one embodiment, the PVA may perform computer stereo vision functions on input from two monocular cameras.

[0199] In at least one embodiment, the PVA may be used to perform dense optical flow. For example, in at least one embodiment, the PVA may process raw RADAR data (e.g., using a 4D Fast Fourier Transform) to provide processed RADAR data. In at least one embodiment, the PVA may be used for time-of-flight depth processing, for example, by processing the raw time-of-flight data to provide processed time-of-flight data.

[0200] In at least one embodiment, DLA may be used to run any type of network for improving control and driving safety, including, for example, but not limited to, a neural network that outputs a measure of confidence for each object detection. In at least one embodiment, confidence may be expressed or interpreted as the probability of each detection compared to other detections or as providing its relative “weight.” In at least one embodiment, confidence allows the system to make further decisions regarding which detections should be considered true positives rather than false positives. For example, in at least one embodiment, the system may set a confidence threshold and consider only detections above the threshold to be true positives. In embodiments where an automatic emergency braking (“AEB”) system is used, a false positive detection may cause the vehicle to automatically apply emergency braking, which is clearly undesirable. In at least one embodiment, a highly confident detection may be considered a trigger for AEB. In at least one embodiment, DLA may run a neural network to regress a confidence value. In at least one embodiment, the neural network may take as its input at least some subset of parameters, such as bounding box dimensions, a ground plane estimate obtained (e.g., from another subsystem), an output from IMU sensor(s) 1666 that correlates with the orientation of the vehicle 1600, distance, and a 3D location estimate of the object obtained from the neural network and / or other sensors (e.g., LIDAR sensor(s) 1664 or RADAR sensor(s) 1660), among others.

[0201] In at least one embodiment, one or more of the SoC(s) 1604 may include data store(s) 1616 (e.g., memory). In at least one embodiment, the data store(s) 1616 may be on-chip memory of the SoC(s) 1604, which may store neural networks to be executed on the GPU(s) 1608 and / or DLA. In at least one embodiment, the data store(s) 1616 may be large enough in capacity to store multiple instances of a neural network for redundancy and safety. In at least one embodiment, the data store(s) 1616 may comprise L2 or L3 cache(s).

[0202] In at least one embodiment, one or more of the SoC(s) 1604 may include number(s) of processor(s) 1610 (e.g., embedded processor(s)). In at least one embodiment, the processor(s) 1610 may include a boot and power management processor, which may be a dedicated processor and subsystem for handling boot power and management functions and related security enforcement. In at least one embodiment, the boot and power management processor may be part of the boot sequence of the SoC(s) 1604 and may provide runtime power management services. In at least one embodiment, the boot power and management processor may provide clock and voltage programming, assist with system low power state transitions, manage thermal and temperature sensors of the SoC(s) 1604, and / or manage the power state of the SoC(s) 1604. In at least one embodiment, each temperature sensor may be implemented as a ring oscillator whose output frequency is proportional to temperature, and the SoC(s) 1604 may use the ring oscillator to detect the temperature of the CPU(s) 1606, the GPU(s) 1608, and / or the accelerator(s) 1614. In at least one embodiment, if the temperature is determined to exceed a threshold, the boot and power management processor may enter a temperature fault routine, place the SoC(s) 1604 in a low power state, and / or place the vehicle 1600 in a chauffeur to safe stop mode (e.g., bring the vehicle 1600 to a safe stop).

[0203] In at least one embodiment, processor(s) 1610 may further include a set of embedded processors that may act as an audio processing engine. In at least one embodiment, the audio processing engine may be an audio subsystem that enables full hardware support for multi-channel audio over multiple interfaces and a wide variety of flexible audio I / O interfaces. In at least one embodiment, the audio processing engine is a dedicated processor core with a digital signal processor with dedicated RAM.

[0204] In at least one embodiment, the processor(s) 1610 may further include an always-on processor engine that may provide the hardware features necessary to support low-power sensor management and wake-up use cases. In at least one embodiment, the always-on processor engine may include, but is not limited to, a processor core, tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I / O controller peripherals, and routing logic.

[0205] In at least one embodiment, the processor(s) 1610 may further include a safety cluster engine, which may include, but is not limited to, a dedicated processor subsystem for handling safety management for automotive applications. In at least one embodiment, the safety cluster engine may include, but is not limited to, two or more processor cores, tightly coupled RAM, supporting peripherals (e.g., timers, interrupt controllers, etc.), and / or routing logic. In safety mode, the two or more cores, in at least one embodiment, may operate in lockstep mode and function as a single core with comparison logic to detect any differences between their operations. In at least one embodiment, the processor(s) 1610 may further include a real-time camera engine, which may include, but is not limited to, a dedicated processor subsystem for handling real-time camera management. In at least one embodiment, the processor(s) 1610 may further include a high dynamic range signal processor, which may include, but is not limited to, an image signal processor, which is a hardware engine that is part of a camera processing pipeline.

[0206] In at least one embodiment, the processor(s) 1610 may include a video image composer, which may be a processing block (e.g., implemented on a microprocessor) that implements video post-processing functions required by a video playback application to produce a final image for a player window. In at least one embodiment, the video image composer may perform lens distortion correction for the wide-angle camera(s) 1670, the surrounding camera(s) 1674, and / or the in-cabin surveillance camera sensor(s). In at least one embodiment, the in-cabin surveillance camera sensor(s) is / are preferably monitored by a neural network running on a separate instance of the SoC(s) 1604 configured to identify events within the cabin and respond accordingly. In at least one embodiment, the in-cabin system may perform lip reading to, but not limited to, activate cellular service, make phone calls, write emails, change vehicle destinations, activate or change the vehicle's infotainment system and settings, or provide voice-activated web surfing. In at least one embodiment, some functions are available to the driver when the vehicle is operating in autonomous mode and are disabled at other times.

[0207] In at least one embodiment, the video image combiner may include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, in at least one embodiment, when motion occurs in the video, noise reduction appropriately weights spatial information and reduces the weight of information provided by adjacent frames. In at least one embodiment, when an image or portion of an image does not contain motion, the temporal noise reduction performed by the video image combiner may use information from previous images to reduce noise in the current image.

[0208] In at least one embodiment, the video image composer may also be configured to perform stereo rectification on the input stereo lens frames. In at least one embodiment, the video image composer may further be used for user interface compositing when the operating system desktop is in use, so that the GPU(s) 1608 are not required to continually render new surfaces. In at least one embodiment, when the GPU(s) 1608 are powered on, active, and performing 3D rendering, the video image composer may be used to offload the GPU(s) 1608 to improve performance and responsiveness.

[0209] In at least one embodiment, one or more of the SoC1604(s) may further include a mobile industry processor interface ("MIPI") camera serial interface, a high-speed interface for receiving video and input from a camera, and / or a video input block that may be used for camera and related pixel input functions. In at least one embodiment, one or more of the SoC1604(s) may further include input / output controller(s), which may be controlled by software and may be used to receive I / O signals that are not committed to a specific role.

[0210] In at least one embodiment, one or more of the SoC(s) 1604 may further include peripherals, audio encoders / decoders ("codecs"), power management, and / or a wide range of peripheral interfaces to enable communication with other devices. The SoC(s) 1604 may be used to process data from cameras (e.g., connected via gigabit multimedia serial links and Ethernet), data from sensors (e.g., LIDAR sensor(s) 1664, RADAR sensor(s) 1660, etc., which may be connected via Ethernet), data from bus 1602 (e.g., vehicle 1600 speed, steering wheel position, etc.), data from GNSS sensor(s) 1658 (e.g., connected via Ethernet or CAN bus), etc. In at least one embodiment, one or more of the SoC(s) 1604 may further include dedicated high performance mass storage controllers, which may include their own DMA engines and may be used to offload the CPU(s) 1606 from routine data management tasks.

[0211] In at least one embodiment, the SoC(s) 1604 may be an end-to-end platform with a flexible architecture spanning levels 3-5 of automation, providing a comprehensive functional safety architecture that leverages and efficiently utilizes computer vision and ADAS techniques for diversity and redundancy, and providing a platform for a flexible and reliable driving software stack, along with deep learning tools. In at least one embodiment, the SoC(s) 1604 may be faster, more reliable, and more energy- and space-efficient than conventional systems. For example, in at least one embodiment, the accelerator(s) 1614, when combined with the CPU(s) 1606, GPU(s) 1608, and data store(s) 1616, may provide a fast and efficient platform for a level 3-5 autonomous vehicle.

[0212] In at least one embodiment, computer vision algorithms may be executed on a CPU, which may be configured using a high-level programming language, such as the C programming language, to perform a wide variety of processing algorithms across a wide variety of visual data. However, in at least one embodiment, CPUs often cannot meet the performance requirements of many computer vision applications, including requirements related to execution time and power consumption. In at least one embodiment, many CPUs are unable to execute complex object detection algorithms in real time, as used in in-vehicle ADAS applications and practical Level 3-5 autonomous vehicles.

[0213] The embodiments described herein allow multiple neural networks to be implemented simultaneously and / or sequentially, with the results being combined together to enable Levels 3-5 autonomous driving functionality. For example, in at least one embodiment, a CNN running on the DLA or a separate GPU (e.g., GPU(s) 1620) can include text and word recognition, enabling the supercomputer to read and understand traffic signs, including signs for which the neural network was not specifically trained. In at least one embodiment, the DLA can further include a neural network that can identify and interpret signs and provide a semantic understanding of the signs, which can then be passed to a route planning module running on the CPU complex.

[0214] In at least one embodiment, multiple neural networks may be run simultaneously for Level 3, 4, or 5 driving. For example, in at least one embodiment, a warning sign consisting of "Caution: Flashing lights indicate icy conditions" along with an electric light may be interpreted independently or collectively by several neural networks. In at least one embodiment, the sign itself may be identified as a traffic sign by a first deployed neural network (e.g., a trained neural network), and the text "Flashing lights indicate icy conditions" may be interpreted by a second deployed neural network, which, when the flashing light is detected, informs the vehicle's route planning software (preferably running on a CPU complex) that an icy condition exists. In at least one embodiment, the flashing light may be identified by running a third deployed neural network over multiple frames, and the third deployed neural network informs the vehicle's route planning software of the presence (or absence) of the flashing light. In at least one embodiment, all three neural networks may run simultaneously, such as within the DLA and / or on the GPU(s) 1608.

[0215] In at least one embodiment, a CNN for facial recognition and vehicle owner identification may use data from the camera sensor to identify the presence of an authorized driver and / or owner of the vehicle 1600. In at least one embodiment, an always-on sensor processing engine may be used to unlock the vehicle when the owner approaches the driver's door and turns on the lights, and in security mode, disable the vehicle when the owner leaves the vehicle. In this manner, the SoC(s) 1604 provide security against theft and / or carjacking.

[0216] In at least one embodiment, a CNN for emergency vehicle detection and identification may detect and identify emergency vehicle sirens using data from microphone 1696. In at least one embodiment, SoC(s) 1604 uses a CNN to classify environmental and urban sounds as well as visual data. In at least one embodiment, a CNN running on the DLA is trained to identify the relative speed at which an emergency vehicle is approaching (e.g., by using the Doppler effect). In at least one embodiment, the CNN may also be trained to identify emergency vehicles specific to the region in which the vehicle is operating, as identified by GNSS sensor(s) 1658. In at least one embodiment, when operating in Europe, the CNN attempts to detect European sirens, and when in the United States, the CNN attempts to identify only North American sirens. In at least one embodiment, when an emergency vehicle is detected, the control program may be used to execute an emergency vehicle safety routine, slow the vehicle, pull over, stop the vehicle, and / or idle the vehicle in conjunction with the ultrasonic sensor(s) 1662 until the emergency vehicle(s) have passed.

[0217] In at least one embodiment, vehicle 1600 may include CPU(s) 1618 (e.g., discrete CPU(s) or dCPU(s)), which may be coupled to SoC(s) 1604 via a high-speed interconnect (e.g., PCIe). In at least one embodiment, CPU(s) 1618 may include, for example, an X86 processor. CPU(s) 1618 may be used to perform any of a variety of functions, including, for example, reconciling potentially inconsistent results between ADAS sensors and SoC(s) 1604 and / or monitoring the status and health of controller(s) 1636 and / or infotainment system on a chip (“infotainment SoC”) 1630.

[0218] In at least one embodiment, vehicle 1600 may include GPU(s) 1620 (e.g., discrete GPU(s) or dGPU(s)), which may be coupled to SoC(s) 1604 via a high-speed interconnect (e.g., NVIDIA's NVLINK). In at least one embodiment, GPU(s) 1620 may provide additional artificial intelligence functionality, such as by running redundant and / or different neural networks, and may be used to train and / or update neural networks based at least in part on input (e.g., sensor data) from sensors of vehicle 1600.

[0219] In at least one embodiment, vehicle 1600 may further include a network interface 1624, which may include, but is not limited to, wireless antenna(s) 1626 (e.g., one or more wireless antenna(s) 1626 for different communication protocols, such as a cellular antenna, a Bluetooth antenna, etc.). In at least one embodiment, network interface 1624 may be used to enable wireless connectivity over the Internet with the cloud (e.g., with server(s) and / or other network devices), with other vehicles, and / or with computing devices (e.g., passenger client devices). In at least one embodiment, to communicate with other vehicles, a direct link may be established between vehicle 160 and the other vehicles and / or an indirect link may be established (e.g., across a network and via the Internet). In at least one embodiment, the direct link may be provided using a vehicle-to-vehicle communication link. The vehicle-to-vehicle communication link may provide vehicle 1600 with information about vehicles in its vicinity (e.g., vehicles in front of, to the sides of, and / or behind vehicle 1600). In at least one embodiment, the aforementioned functionality may be part of the cooperative adaptive cruise control functionality of vehicle 1600.

[0220] In at least one embodiment, the network interface 1624 may include an SoC that provides modulation and demodulation functionality and enables the controller(s) 1636 to communicate over a wireless network. In at least one embodiment, the network interface 1624 may include a radio frequency front end for up-conversion from baseband to radio frequency and down-conversion from radio frequency to baseband. In at least one embodiment, the frequency conversion may be performed in any technically feasible manner. For example, the frequency conversion may be performed through well-known processes and / or using a super-heterodyne process. In at least one embodiment, the radio frequency front end functionality may be provided by a separate chip. In at least one embodiment, the network interface may include wireless functionality for communicating via LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-Wave, ZigBee, LoRaWAN, and / or other wireless protocols.

[0221] In at least one embodiment, vehicle 1600 may further include data store(s) 1628, which may include, but are not limited to, off-chip (e.g., not on SoC(s) 1604) storage. In at least one embodiment, data store(s) 1628 may include one or more storage elements, including, but not limited to, RAM, SRAM, dynamic random access memory (“DRAM”), video random-access memory (“VRAM”), flash, hard disk, and / or other components and / or devices capable of storing at least one bit of data.

[0222] In at least one embodiment, vehicle 1600 may further include GNSS sensor(s) 1658 (e.g., GPS and / or assisted GPS sensors) to assist in mapping, perception, occupancy grid generation, and / or route planning functions. In at least one embodiment, any number of GNSS sensors 1658 may be used, including, for example, but not limited to, a GPS using a USB connector with an Ethernet-to-serial (e.g., RS-232) bridge.

[0223] In at least one embodiment, vehicle 1600 may further include RADAR sensor(s) 1660. RADAR sensor(s) 1660 may be used by vehicle 1600 for long-range vehicle detection, even in darkness and / or severe weather conditions. In at least one embodiment, the RADAR functional safety level may be ASIL B. RADAR sensor(s) 1660 may use CAN and / or bus 1602 for control (e.g., to transmit data generated by RADAR sensor(s) 1660) and to access object tracking data, in some instances, along with access to Ethernet for accessing raw data. In at least one embodiment, a wide variety of RADAR sensor types may be used. For example, without limitation, RADAR sensor(s) 1660 may be suitable for forward, rear, and side RADAR use. In at least one embodiment, one or more of the RADAR sensor(s) 1660 are pulse-Doppler RADAR sensor(s).

[0224] In at least one embodiment, the RADAR sensor(s) 1660 may include different configurations, such as long-range with a narrow field of view, short-range with a wide field of view, and short-range lateral coverage. In at least one embodiment, the long-range RADAR may be used for adaptive cruise control functionality. In at least one embodiment, the long-range RADAR system may provide a wide field of view achieved by two or more independent scans, such as within a 250 meter range. In at least one embodiment, the RADAR sensor(s) 1660 may help distinguish between static and moving objects and may be used by the ADAS system 1638 for emergency braking assistance and forward collision warning. The sensor(s) 1660 included in a long-range RADAR system may include, but are not limited to, multiple (e.g., six or more) fixed RADAR antennas, as well as monostatic and multimodal RADARs with high-speed CAN and FlexRay interfaces. In at least one embodiment, where there are six antennas, the central four antennas may create a focused beam pattern designed to record the surroundings of vehicle 1600 at higher speeds with minimal interference from traffic in adjacent lanes. In at least one embodiment, the other two antennas may increase the field of view, which may allow for quick detection of vehicles entering or exiting vehicle 1600's lane.

[0225] In at least one embodiment, the medium-range RADAR system may include, by way of example, a range of up to 160 meters (forward) or 80 meters (rearward) and a field of view of up to 42 degrees (forward) or 150 degrees (rearward). In at least one embodiment, the short-range RADAR system may include any number of RADAR sensors 1660 designed to be mounted on either end of a rear bumper, without limitation. When mounted on either end of a rear bumper, in at least one embodiment, the RADAR sensor system may create two beams that constantly monitor blind spots behind and adjacent to the vehicle. In at least one embodiment, the short-range RADAR system may be used in an ADAS system 1638 for blind spot detection and / or lane change assistance.

[0226] In at least one embodiment, the vehicle 1600 may further include ultrasonic sensor(s) 1662. The ultrasonic sensor(s) 1662, which may be positioned at the front, rear, and / or sides of the vehicle 1600, may be used for parking assistance and / or to create and update an occupancy grid. In at least one embodiment, a variety of ultrasonic sensor(s) 1662 may be used, and different ultrasonic sensor(s) 1662 may be used for different detection ranges (e.g., 2.5 m, 4 m). In at least one embodiment, the ultrasonic sensor(s) 1662 may operate at a functional safety level of ASIL B.

[0227] In at least one embodiment, the vehicle 1600 may include one or more LIDAR sensors 1664. The LIDAR sensor(s) 1664 may be used for object and pedestrian detection, emergency braking, collision avoidance, and / or other functions. In at least one embodiment, the vehicle 1600 may include multiple LIDAR sensors 1664 (e.g., two, four, six, etc.), and the LIDAR sensors 1664 may use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).

[0228] In at least one embodiment, the LIDAR sensor(s) 1664 may be capable of providing a list of objects and their distances for a 360-degree field of view. In at least one embodiment, commercially available LIDAR sensor(s) 1664 may have an advertised range of approximately 100 meters, with an accuracy of 2 cm to 3 cm, and support for a 100 Mbps Ethernet connection, for example. In at least one embodiment, one or more non-protruding LIDAR sensors 1664 may be used. In such an embodiment, the LIDAR sensor(s) 1664 may be implemented as small devices that may be integrated into the front, rear, sides, and / or corners of the vehicle 1600. In at least one embodiment, the LIDAR sensor(s) 1664 may provide a horizontal field of view of up to 120 degrees and a vertical field of view of 35 degrees, with a range of 200 meters even for low-reflectivity objects. In at least one embodiment, the forward mounted LIDAR sensor(s) 1664 may be configured for a horizontal field of view between 45 degrees and 135 degrees.

[0229] In at least one embodiment, LIDAR technology such as 3D flash LIDAR may also be used. 3D flash LIDAR uses a laser flash as a transmitting source to illuminate the surroundings of the vehicle 1600 up to approximately 200 meters. In at least one embodiment, the flash LIDAR unit may include, but is not limited to, a receptor that records the transit time of the laser pulse and the reflected light on each pixel, which corresponds to the range from the vehicle 1600 to the object. In at least one embodiment, flash LIDAR may enable a highly accurate, distortion-free image of the surroundings to be generated with each laser flash. In at least one embodiment, four flash LIDAR sensors may be deployed, one on each side of the vehicle 1600. In at least one embodiment, the 3D flash LIDAR system may include, but is not limited to, a solid-state 3D staring array LIDAR camera (e.g., a non-scanning LIDAR device) with no moving parts other than a fan. In at least one embodiment, the flash LIDAR device(s) may use 5 nanosecond Class I (eye-safe) laser pulses per frame and may capture reflected laser light in the form of a 3D range point cloud and co-registered intensity data.

[0230] In at least one embodiment, the vehicle may further include IMU sensor(s) 1666. In at least one embodiment, the IMU sensor(s) 1666 may be located at the center of the rear axle of the vehicle 1600. In at least one embodiment, the IMU sensor(s) 1666 may include, for example, but not limited to, accelerometer(s), magnetometer(s), gyroscope(s), magnetic compass(es), and / or other sensor types. In at least one embodiment, such as in a 6-axis application, the IMU sensor(s) 1666 may include, but are not limited to, an accelerometer and a gyroscope. In at least one embodiment, such as in a 9-axis application, the IMU sensor(s) 1666 may include, but are not limited to, an accelerometer, a gyroscope, and a magnetometer.

[0231] In at least one embodiment, the IMU sensor(s) 1666 may be implemented as a compact, high-performance GPS-Aided Inertial Navigation System ("GPS / INS") that combines micro-electro-mechanical systems ("MEMS") inertial sensors, a highly sensitive GPS receiver, and advanced Kalman filtering algorithms to provide estimates of position, velocity, and attitude. In at least one embodiment, the IMU sensor(s) 1666 may enable the vehicle 1600 to estimate orientation without requiring input from a magnetic sensor by directly observing changes in velocity and correlating that from GPS to the IMU sensor(s) 1666. In at least one embodiment, the IMU sensor(s) 1666 and the GNSS sensor(s) 1658 may be combined in a single, integrated unit.

[0232] In at least one embodiment, vehicle 1600 may include microphone(s) 1696 positioned in and / or around vehicle 1600. In at least one embodiment, microphone(s) 1696 may be used for, among other things, emergency vehicle detection and identification.

[0233] In at least one embodiment, vehicle 1600 may further include any number of camera types, including stereo camera(s) 1668, wide-angle camera(s) 1670, infrared camera(s) 1672, surrounding camera(s) 1674, long-range camera(s) 1698, mid-range camera(s) 1676, and / or other camera types. In at least one embodiment, cameras may be used to capture image data around the entire perimeter of vehicle 1600. In at least one embodiment, the types of cameras used depend on vehicle 1600. In at least one embodiment, any combination of camera types may be used to provide the required coverage around vehicle 1600. In at least one embodiment, the number of cameras may vary depending on the embodiment. For example, in at least one embodiment, vehicle 1600 may include six cameras, seven cameras, ten cameras, twelve cameras, or another number of cameras. The cameras may support, by way of example and not limitation, Gigabit Multimedia Serial Link ("GMSL") and / or Gigabit Ethernet. In at least one embodiment, each of the camera(s) is described in more detail herein above with respect to Figures 16A and 16B.

[0234] In at least one embodiment, vehicle 1600 may further include vibration sensor(s) 1642. In at least one embodiment, vibration sensor(s) 1642 may measure vibration of a component of vehicle 1600, such as an axle(s). For example, in at least one embodiment, a change in vibration may indicate a change in the road surface. In at least one embodiment, when two or more vibration sensors 1642 are used, the difference in vibration may be used to determine the amount of friction or slippage of the road surface (e.g., when the difference in vibration is between a powered axle and a free-spinning axle).

[0235] In at least one embodiment, vehicle 1600 may include an ADAS system 1638. ADAS system 1638 may include, but is not limited to, an SoC in some instances. In at least one embodiment, the ADAS system 1638 may include, without limitation, any number and combination of autonomous / adaptive / automatic cruise control ("ACC") systems, cooperative adaptive cruise control ("CACC") systems, forward crash warning ("FCW") systems, automatic emergency braking ("AEB") systems, lane departure warning ("LDW") systems, lane keep assist ("LKA") systems, blind spot warning ("BSW") systems, rear cross-traffic warning ("RCTW") systems, collision warning ("CW") systems, lane centering ("LC") systems, and / or other systems, features, and / or functionality.

[0236] In at least one embodiment, the ACC system may use RADAR sensor(s) 1660, LIDAR sensor(s) 1664, and / or any number of cameras. In at least one embodiment, the ACC system may include a longitudinal ACC system and / or a lateral ACC system. In at least one embodiment, the longitudinal ACC system monitors and controls the distance to the vehicle immediately preceding the vehicle 1600 and automatically adjusts the speed of the vehicle 1600 to maintain a safe distance from the vehicle ahead. In at least one embodiment, the lateral ACC system enforces distance maintenance and advises the vehicle 1600 to change lanes when necessary. In at least one embodiment, the lateral ACC is related to other ADAS applications, such as LC and CW.

[0237] In at least one embodiment, the CACC system uses information from other vehicles, which may be received via a wireless link or indirectly, via a network connection (e.g., via the Internet), from other vehicles via network interface 1624 and / or wireless antenna(s) 1626. In at least one embodiment, the direct link may be provided by a vehicle-to-vehicle ("V2V") communication link, and the indirect link may be provided by an infrastructure-to-vehicle ("I2V") communication link. Generally, V2V communication concepts provide information about an immediate preceding vehicle (e.g., a vehicle immediately preceding vehicle 1600 and in the same lane), while I2V communication concepts provide information about traffic further ahead. In at least one embodiment, the CACC system may include either or both I2V and V2V information sources. In at least one embodiment, with information about vehicles in front of vehicle 1600, the CACC system can be more reliable, which has the potential to improve traffic flow and reduce congestion on roads.

[0238] In at least one embodiment, the FCW system is designed to alert the driver of hazards so that the driver can take corrective action. In at least one embodiment, the FCW system uses a front-facing camera and / or RADAR sensor(s) 1660 coupled to a dedicated processor, DSP, FPGA, and / or ASIC electrically coupled to driver feedback, such as a display, speaker, and / or vibration components. In at least one embodiment, the FCW system may provide warnings in the form of an audible, visual warning, vibration, and / or quick brake pulse.

[0239] In at least one embodiment, the AEB system may detect an imminent forward collision with another vehicle or other object and automatically apply the brakes if the driver does not take corrective action within specified time or distance parameters. In at least one embodiment, the AEB system may use a front-facing camera(s) and / or a RADAR sensor(s) 1660 coupled to a dedicated processor, DSP, FPGA, and / or ASIC. In at least one embodiment, when the AEB system detects a hazard, the AEB system typically first alerts the driver to take corrective action to avoid the collision; if the driver does not take corrective action, the AEB system may automatically apply the brakes to prevent or at least mitigate the impact of the predicted collision. In at least one embodiment, the AEB system may include techniques such as dynamic brake support and / or pre-crash braking.

[0240] In at least one embodiment, the LDW system provides visual, audible, and / or tactile warnings, such as vibration of the steering wheel or seat, to alert the driver when the vehicle 1600 crosses a lane marker. In at least one embodiment, the LDW system does not activate when the driver indicates an intentional lane departure by activating a turn signal. In at least one embodiment, the LDW system may use a front-facing camera coupled to a dedicated processor, DSP, FPGA, and / or ASIC electrically coupled to driver feedback, such as a display, speaker, and / or vibration components. In at least one embodiment, the LKA system is a variation of the LDW system. The LKA system provides steering inputs or brake control to correct the vehicle 1600 if the vehicle 1600 begins to drift out of its lane.

[0241] In at least one embodiment, the BSW system detects vehicles in the vehicle's blind spot and warns the driver about the vehicles. In at least one embodiment, the BSW system may provide visual, audible, and / or tactile alerts to indicate that it is unsafe to merge or change lanes. In at least one embodiment, the BSW system may provide an additional warning when the driver uses a turn signal. In at least one embodiment, the BSW system may use rear-facing camera(s) and / or RADAR sensor(s) 1660 coupled to a dedicated processor, DSP, FPGA, and / or ASIC electrically coupled to driver feedback, such as a display, speaker, and / or vibration components.

[0242] In at least one embodiment, the RCTW system may provide visual, audible, and / or tactile notifications when an object is detected outside the range of the rear camera when the vehicle 1600 is reversing. In at least one embodiment, the RCTW system includes an AEB system to ensure vehicle brakes are applied to avoid a crash. In at least one embodiment, the RCTW system may use one or more rear RADAR sensors 1660 coupled to a dedicated processor, DSP, FPGA, and / or ASIC electrically coupled to driver feedback, such as a display, speaker, and / or vibration components.

[0243] In at least one embodiment, conventional ADAS systems may be prone to false positive results, which can be annoying and distracting to the driver, but this is usually not a major issue because conventional ADAS systems alert the driver, allowing the driver to determine whether a safety condition truly exists and act accordingly. In at least one embodiment, the vehicle 1600 itself determines whether to follow the result from the primary computer (e.g., the first controller 1636) or the result from the secondary computer (e.g., the second controller 1636) in the case of conflicting results. For example, in at least one embodiment, the ADAS system 1638 may be a backup and / or secondary computer for providing perception information to a backup computer rationality module. In at least one embodiment, a backup computer rationality monitor may run redundant and diverse software on hardware components to detect perception and dynamic driving task impairments. In at least one embodiment, output from the ADAS system 1638 may be provided to a supervisory MCU. In at least one embodiment, if the output from the primary computer and the output from the secondary computer conflict, the supervising MCU determines how to reconcile the conflict to ensure safe operation.

[0244] In at least one embodiment, the primary computer may be configured to provide the overseer MCU with a reliability score indicating the primary computer's reliability in a selected outcome. In at least one embodiment, if the reliability score exceeds a threshold, the overseer MCU may follow the primary computer's instructions regardless of whether the secondary computers provide conflicting or inconsistent results. In at least one embodiment, if the reliability score does not meet the threshold, and if the primary and secondary computers exhibit different (e.g., conflicting) results, the overseer MCU may arbitrate between the computers to determine an appropriate outcome.

[0245] In at least one embodiment, the supervisory MCU can be configured to run neural network(s) trained and configured to determine conditions under which the secondary computer will provide a false alarm based at least in part on outputs from the primary and secondary computers. In at least one embodiment, the neural network(s) in the supervisory MCU can learn when the output of the secondary computer can be trusted and when it cannot be trusted. For example, in at least one embodiment, when the secondary computer is a RADAR-based FCW system, the neural network(s) in the supervisory MCU can learn when the FCW system identifies a metal object that is not actually a hazard, such as a drain grate or manhole cover, which triggers an alarm. In at least one embodiment, when the secondary computer is a camera-based LDW system, the neural network in the supervisory MCU can learn to disable LDW when a bicyclist or pedestrian is present and lane departure is actually the safest maneuver. In at least one embodiment, the supervising MCU may include at least one of a DLA or a GPU suitable for running neural network(s) along with associated memory. In at least one embodiment, the supervising MCU may comprise and / or be included as a component of SoC1604(s).

[0246] In at least one embodiment, the ADAS system 1638 may include a secondary computer that implements ADAS functionality using traditional rules of computer vision. In at least one embodiment, the secondary computer may use traditional computer vision rules (if-then), and the presence of neural network(s) in the supervisory MCU may improve reliability, safety, and performance. For example, in at least one embodiment, diverse implementations and intentional non-identity make the overall system more fault-tolerant, particularly to failures caused by software (or software-hardware interface) functionality. For example, in at least one embodiment, if there is a software bug or error in software running on the primary computer and non-identical software code running on the secondary computer provides the same overall result, the supervisory MCU may have higher confidence that the overall result is correct and that a bug in the software or hardware on the primary computer did not cause a critical error.

[0247] In at least one embodiment, the output of the ADAS system 1638 can be fed to a perception block of the primary computer and / or a dynamic driving task block of the primary computer. For example, in at least one embodiment, if the ADAS system 1638 indicates a frontal crash warning due to an upcoming object, the perception block can use this information when identifying the object. In at least one embodiment, the secondary computer can have its own neural network trained as described herein, thus reducing the risk of false positives.

[0248] In at least one embodiment, vehicle 1600 may further include an infotainment SoC 1630 (e.g., an in-vehicle infotainment system (IVI)). Although shown and described as an SoC, infotainment system 1630, in at least one embodiment, may not be an SoC and may include, without limitation, two or more separate components. In at least one embodiment, infotainment SoC 1630 may include, but is not limited to, a combination of hardware and software that may be used to provide audio (e.g., music, personal digital assistant, navigation instructions, news, radio, etc.), video (e.g., TV, movies, streaming, etc.), telephony (e.g., hands-free calling), network connectivity (e.g., LTE, Wi-Fi, etc.), and / or information services (e.g., navigation systems, rear parking assist, wireless data systems, vehicle-related information such as fuel level, total mileage, brake fuel level, oil level, door opening / closing, air filter information, etc.) to vehicle 1600. For example, infotainment SoC 1630 may include a radio, a disc player, a navigation system, a video player, USB and Bluetooth connectivity, a car computer, in-car entertainment, Wi-Fi, steering wheel audio controls, hands-free voice control, a heads-up display (“HUD”), an HMI display 1634, telematics devices, a control panel (e.g., for controlling and / or interacting with various components, features, and / or systems), and / or other components. In at least one embodiment, infotainment SoC 1630 may further be used to provide information (e.g., visual and / or auditory) to a user(s) of the vehicle, such as information from an ADAS system 1638, autonomous driving information such as planned vehicle maneuvers, trajectories, surrounding environment information (e.g., intersection information, vehicle information, road information, etc.), and / or other information.

[0249] In at least one embodiment, infotainment SoC 1630 may include any amount and type of GPU functionality. In at least one embodiment, infotainment SoC 1630 may communicate with other devices, systems, and / or components of vehicle 1600 via bus 1602 (e.g., CAN bus, Ethernet, etc.). In at least one embodiment, infotainment SoC 1630 may be coupled to a supervisory MCU such that the infotainment system's GPU may perform some self-driving functions if primary controller(s) 1636 (e.g., vehicle 1600's primary and / or backup computer) fails. In at least one embodiment, infotainment SoC 1630 may place vehicle 1600 in a driver-safety shutdown mode, as described herein.

[0250] In at least one embodiment, vehicle 1600 may further include an instrument cluster 1632 (e.g., a digital dashboard, an electronic instrument cluster, a digital instrument panel, etc.). In at least one embodiment, instrument cluster 1632 may include, but is not limited to, a controller and / or a supercomputer (e.g., a separate controller or supercomputer). In at least one embodiment, instrument cluster 1632 may include any number and combination of instrumentation sets, such as, but not limited to, a speedometer, fuel level, oil pressure, a tachometer, an odometer, direction indicators, a shift lever position indicator, seat belt warning light(s), parking brake warning light(s), engine malfunction light(s), supplemental restraint system (e.g., airbag) information, lighting control, safety system control, navigation information, etc. In some instances, information may be displayed and / or shared between infotainment SoC 1630 and instrument cluster 1632. In at least one embodiment, the instrument cluster 1632 may be included as part of the infotainment SoC 1630, or vice versa.

[0251] Inference and / or training logic 715 is used to perform inference and / or training operations associated with one or more embodiments. More details regarding inference and / or training logic 715 are provided below in conjunction with Figures 7A and / or 7B. In at least one embodiment, inference and / or training logic 715 may be used in the system of Figure 16A for inference or prediction operations based at least in part on weight parameters calculated using neural network training operations, neural network functionality and / or architecture, or neural network use cases described herein.

[0252] 16B is a diagram of a system 1676 for communication between cloud-based server(s) and the autonomous vehicle 1600 of FIG. 16A , according to at least one embodiment. In at least one embodiment, the system 1676 may include, but is not limited to, server(s) 1678, network(s) 1690, and any number and type of vehicles, including vehicle 1600. In at least one embodiment, the server(s) 1678 may include, but is not limited to, multiple GPUs 1684(A)-1684(H) (collectively referred to herein as GPUs 1684), PCIe switches 1682(A)-1682(D) (collectively referred to herein as PCIe switch 1682), and / or CPUs 1680(A)-1680(B) (collectively referred to herein as CPUs 1680). The GPUs 1684, CPUs 1680, and PCIe switches 1682 may be interconnected with a high-speed interconnect, such as, but not limited to, an NVLink interface 1688 and / or PCIe connections 1686 developed by NVIDIA. In at least one embodiment, the GPUs 1684 are connected via an NVLink and / or NVSwitch SoC, and the GPUs 1684 and PCIe switches 1682 are connected via a PCIe interconnect. In at least one embodiment, eight GPUs 1684, two CPUs 1680, and four PCIe switches 1682 are shown, but this is not intended to be limiting. In at least one embodiment, each of the server(s) 1678 may include any number of GPUs 1684, CPUs 1680, and / or PCIe switches 1682 in any combination, including, but not limited to, For example, in at least one embodiment, server(s) 1678 may each include 8, 16, 32, and / or more GPUs 1684.

[0253] In at least one embodiment, server(s) 1678 may receive image data from the vehicle over network(s) 1690 representing images showing unexpected or changed road conditions, such as recently begun road construction. In at least one embodiment, server(s) 1678 may transmit neural network 1692, updated neural network 1692, and / or map information 1694, including, but not limited to, information regarding traffic and road conditions, to the vehicle over network(s) 1690. In at least one embodiment, updates to map information 1694 may include updates to HD map 1622, such as, but not limited to, information regarding construction sites, potholes, detours, flooding, and / or other obstacles. In at least one embodiment, neural network 1692, updated neural network 1692, and / or map information 1694 may result from new training and / or experience represented in data received from any number of vehicles in the environment and / or based at least in part on training performed at a data center (e.g., using server(s) 1678 and / or other servers).

[0254] In at least one embodiment, server(s) 1678 may be used to train a machine learning model (e.g., a neural network) based at least in part on the training data. In at least one embodiment, the training data may be generated by the vehicle and / or generated in a simulation (e.g., using a game engine). In at least one embodiment, any amount of the training data may be tagged and / or subjected to other preprocessing (e.g., if the associated neural network benefits from supervised learning). In at least one embodiment, any amount of the training data may not be tagged and / or preprocessed (e.g., if the associated neural network does not require supervised learning). In at least one embodiment, once the machine learning model is trained, it may be used by the vehicle (e.g., transmitted to the vehicle via network(s) 1690) and / or used by server(s) 1678 to remotely monitor the vehicle.

[0255] In at least one embodiment, server(s) 1678 may receive data from vehicles and apply the data to state-of-the-art real-time neural networks for real-time intelligent inference. In at least one embodiment, server(s) 1678 may include deep learning supercomputers and / or special-purpose AI computers powered by GPU(s) 1684, such as the DGX and DGX Station machines developed by NVIDIA. However, in at least one embodiment, server(s) 1678 may include a deep learning infrastructure using CPU-powered data centers.

[0256] In at least one embodiment, the deep learning infrastructure of server(s) 1678 may be capable of fast real-time inference and may use that capability to assess and verify the health of processors, software, and / or associated hardware in vehicle 1600. For example, in at least one embodiment, the deep learning infrastructure may receive periodic updates from vehicle 1600, such as a series of images and / or objects that vehicle 1600 has located in the series of images (e.g., via computer vision and / or other machine learning object classification techniques). In at least one embodiment, the deep learning infrastructure may run its own neural network to identify objects and compare them to the objects identified by vehicle 1600; if the results do not match and the deep learning infrastructure concludes that the AI ​​in vehicle 1600 has failed, server(s) 1678 may send a signal to vehicle 1600 instructing the vehicle's 1600 fail-safe computer to assume control, notify passengers, and complete a safe parking maneuver.

[0257] In at least one embodiment, server(s) 1678 may include GPU(s) 1684 and one or more programmable inference accelerators (e.g., NVIDIA's TensorRT3). In at least one embodiment, the combination of GPU-powered servers and inference acceleration may enable real-time response. In at least one embodiment, servers powered by CPUs, FPGAs, and other processors may be used for inference, such as when performance is less critical. In at least one embodiment, inference and / or training logic 715 is used to implement one or more embodiments. More details regarding the inference and / or training logic 715 are provided below in conjunction with FIG. 7A and / or FIG. 7B.

[0258] Other variations are within the scope of the present disclosure. Thus, while the disclosed techniques are susceptible to various modifications and alternative constructions, several illustrative examples thereof have been shown in the drawings and described above in detail. It should be understood, however, that there is no intention to limit the disclosure to the particular disclosed form or forms, but on the contrary, the intention is to cover all modifications, alternative constructions, and equivalents falling within the spirit and scope of the disclosure as defined in the appended claims.

[0259] The use of the terms "a," "an," and "the," and similar referents in the context of describing the disclosed embodiments (particularly in the context of the claims that follow) should be construed to cover both the singular and the plural and not as defining terms, unless otherwise stated herein or clearly contradicted by context. The terms "comprising," "having," "including," and "containing" should be construed as open-ended terms (meaning "including, but not limited to"), unless otherwise indicated. The term "connected," when unmodified and referring to a physical connection, should be construed as partially or completely contained within, attached to, or joined to one another, even if there is something intervening. Reciting ranges of values ​​herein is merely intended to serve as a shorthand method of individually referring to each separate value falling within the range, unless otherwise stated herein and unless each separate value is incorporated into the specification as if it were individually recited herein. Use of the term "set" (e.g., "set of items") or "subset" should be construed as a non-empty collection comprising one or more members, unless otherwise stated or contradicted by context. Furthermore, unless otherwise stated or contradicted by context, the term "subset" of a corresponding set does not necessarily refer to a strict subset of the corresponding set; a subset and a corresponding set may be equivalent.

[0260] Connective language, such as phrases of the form "at least one of A, B, and C" or "at least one of A, B, and C," is understood in the context in which it is generally used to indicate that an item, term, etc. may be either A or B or C, or any non-empty subset of the set A, B, and C, unless specifically stated otherwise or clearly contradicted by context. For example, in the illustrative example of a set having three members, the connective phrases "at least one of A, B, and C" and "at least one of A, B, and C" refer to any of the following sets: {A}, {B}, {C}, {A, B}, {A, C}, {B, C}, {A, B, C}. Thus, such connective language does not generally imply that some embodiments require that at least one of A, at least one of B, and at least one of C each be present. Further, unless stated otherwise or negated by context, the term "plurality" refers to a plurality (e.g., "a plurality of items" refers to multiple items). A plurality is at least two items, but may be more when so indicated either explicitly or by context. Further, unless stated otherwise or otherwise clear from context, the phrase "based on" means "based at least in part on," and not "based only on."

[0261] The operations of processes described herein may be performed in any suitable order unless otherwise stated herein or otherwise clearly contradicted by context. In at least one embodiment, processes such as those described herein (or variations and / or combinations thereof) are performed under the control of one or more computer systems comprised of executable instructions and implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) that collectively execute on one or more processors, by hardware, or by a combination thereof. In at least one embodiment, the code is stored on a computer-readable storage medium, e.g., in the form of a computer program comprising a plurality of instructions executable by one or more processors. In at least one embodiment, the computer-readable storage medium is a non-transitory computer-readable storage medium that excludes transitory signals (e.g., propagating transitory electrical or electromagnetic transmissions), but includes non-transitory data storage circuitry (e.g., buffers, caches, and queues) within a transceiver of a transitory signal. In at least one embodiment, the code (e.g., executable code or source code) is stored on a set of one or more non-transitory computer-readable storage media that store executable instructions (or have other memory for storing executable instructions) that, when executed by (i.e., as a result of being executed by) one or more processors of a computer system, cause the computer system to perform the operations described herein. The set of non-transitory computer-readable storage media, in at least one embodiment, comprises a plurality of non-transitory computer-readable storage media, wherein one or more individual non-transitory storage media of the plurality of non-transitory computer-readable storage media do not contain all of the code, but the plurality of non-transitory computer-readable storage media collectively store all of the code.In at least one embodiment, the executable instructions are executed such that different instructions are executed by different processors, e.g., a non-transitory computer-readable storage medium stores the instructions, a main central processing unit ("CPU") executes some of the instructions, and a graphics processing unit ("GPU") executes other instructions. In at least one embodiment, different components of a computer system have separate processors, with different processors executing different subsets of the instructions.

[0262] Thus, in at least one embodiment, a computer system is configured to implement one or more services that, singly or collectively, perform the operations of the processes described herein, and such a computer system is configured with applicable hardware and / or software that enables the performance of the operations. Further, a computer system that implements at least one embodiment of the present disclosure is a single device, and in another embodiment, is a distributed computer system comprising multiple devices that operate in different ways, such that a distributed computer system performs the operations described herein, and such that no single device performs all of the operations.

[0263] The use of any illustrative or exemplary language (e.g., "such as") provided herein is intended only to further clarify embodiments of the disclosure and does not limit the scope of the disclosure unless otherwise stated. No language in the specification should be construed as indicating any non-claimed element as essential to the practice of the disclosure.

[0264] All references cited herein, including publications, patent applications, and patents, are hereby incorporated by reference to the same extent as if each reference was individually and specifically indicated to be incorporated by reference and were set forth in its entirety herein.

[0265] In the specification and claims, the terms "coupled" and "connected," along with their derivatives, may be used. It should be understood that these terms may not be intended as synonyms for each other. Rather, in particular instances, "connected" or "coupled" may be used to indicate that two or more elements are in direct or indirect physical or electrical contact with each other. "Coupled" may also mean that two or more elements are not in direct contact with each other, but yet still cooperate or interact with each other.

[0266] Unless specifically stated otherwise, it will be appreciated that throughout the specification, terms such as "processing," "computing," "calculating," or "determining" refer to the actions and / or processes of a computer or computing system, or similar electronic computing device, that manipulate and / or transform data represented as physical quantities, such as electronic quantities, in the computing system's registers and / or memory into other data similarly represented as physical quantities in the computing system's memory, registers, or other such information storage, transmission, or display device.

[0267] Similarly, the term "processor" may refer to any device, or portion of a device, that processes electronic data from registers and / or memory and converts the electronic data into other electronic data that may be stored in registers and / or memory. As a non-limiting example, a "processor" may be a CPU or GPU. A "computing platform" may comprise one or more processors. As used herein, a "software" process may include software and / or hardware entities that perform work over time, such as tasks, threads, and intelligent agents. Also, each process may refer to multiple processes for executing instructions serially or in parallel, continuously or intermittently. The terms "system" and "method" are used interchangeably herein, provided that a system may embody one or more methods and a method may be considered a system.

[0268] As used herein, references may be made to acquiring, obtaining, receiving, or inputting analog or digital data into a subsystem, computer system, or computer-implemented machine. Acquiring, obtaining, receiving, or inputting analog and digital data may be accomplished in various ways, such as by receiving data as parameters of a function call or a call to an application programming interface. In some implementations, the process of acquiring, obtaining, receiving, or inputting analog or digital data may be accomplished by transferring data over a serial or parallel interface. In other implementations, the process of acquiring, obtaining, receiving, or inputting analog or digital data may be accomplished by transferring data over a computer network from a providing entity to an acquiring entity. References may also be made to providing, outputting, transmitting, sending, or presenting analog or digital data. In various instances, the process of providing, outputting, transmitting, sending, or presenting analog or digital data may be accomplished by transferring data as input or output parameters of a function call, an application programming interface, or an inter-process communication mechanism.

[0269] While the above description describes example implementations of the described techniques, other architectures may be used to implement the described functionality and are intended to be within the scope of this disclosure. Additionally, while a specific distribution of responsibilities has been defined above for purposes of explanation, various functions and responsibilities may be distributed and divided in different ways depending on the circumstances.

[0270] Furthermore, although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter claimed in the appended claims is not necessarily limited to the specific features or acts described. Rather, the specific features and acts are disclosed as example forms of implementing the claims.

Claims

1. using an image containing a representation of a human face, identifying a set of facial landmarks in said image; determining eye state information indicative of whether the person's eyes are open or closed from at least a portion of the image; determining head pose information for the person based at least in part on the facial landmarks; determining a set of eyeblink parameters for the person over a recent time period based at least in part on the set of facial landmarks, the head pose information, and the eye state information; using the eye state information to determine blink frequency information for the recent time period; providing the set of eyeblink parameters as input to a first temporal network to generate a first drowsiness prediction for the person; providing the eye blink rate information as an input to a second temporal network to generate a second drowsiness prediction for the person; generating an overall somnolence determination for the person based at least in part on the first somnolence prediction and the second somnolence prediction; 11. A computer-implemented method comprising:

2. detecting a portion of the image corresponding to the face of the person; identifying the set of facial landmarks from the portion of the image; The computer-implemented method of claim 1 , further comprising:

3. using a subset of the facial landmarks to identify one or more eye regions to be used to determine the eye state information. The computer-implemented method of claim 1 , further comprising:

4. 2. The computer-implemented method of claim 1, wherein the first temporal network and the second temporal network are long short-term memory (LSTM) networks, and the first somnolence prediction and the second somnolence prediction generated by the LSTM networks correspond to Karolinska Sleepiness Scale (KSS) values.

5. using the image to determine the identity of the person; identifying an eyeblink profile for the person, the eyeblink profile being indicative of one or more eyeblink behaviors of the person; generating at least the first drowsiness prediction further based on data about the one or more eye blink behaviors; The computer-implemented method of claim 1 , further comprising:

6. The computer-implemented method of claim 1 , wherein the set of eyeblink parameters includes at least eyeblink amplitude, eyeblink duration, or eyeblink velocity.

7. performing a transformation of the set of facial landmarks based at least in part on the head pose information to adjust the orientation of the set of facial landmarks. The computer-implemented method of claim 1 , further comprising:

8. The computer-implemented method of claim 1 , wherein the eye state information is determined from one or more eye portions of the image and is independent of the set of facial landmarks.

9. The computer-implemented method of claim 1 , wherein at least a subset of the set of eyeblink parameters is determined using aspect ratio information calculated from the set of facial landmarks.

10. a face detection network for identifying a portion of an image corresponding to a human face; a facial landmark detection network for identifying a plurality of facial landmarks present in the portion of the image corresponding to the face; an eye state network for determining an eye state of the person from the image, the eye state corresponding to at least open or closed; and a first temporal network for generating a first drowsiness prediction for the person based at least in part on the set of eyeblink parameters determined from the plurality of facial landmarks; a second temporal network for generating a second drowsiness prediction for the person based at least in part on the blink rate determined using the eye state; and A system comprising:

11. a facial recognition module for determining an identity of the person, an eye blink profile to be used to locate an eye blink profile indicative of one or more eye blink behaviors of the person for use in generating at least the first drowsiness prediction; The system of claim 10 further comprising:

12. 11. The system of claim 10, wherein the first temporal network and the second temporal network are long short-term memory (LSTM) networks, and the first drowsiness prediction and the second drowsiness prediction generated by the LSTM networks correspond to Karolinska Sleepiness Scale (KSS) values.

13. a head pose determination network for determining head pose information for the person using the plurality of facial landmarks, the first temporal network for performing a transformation of the plurality of facial landmarks based at least in part on the head pose information to adjust an orientation of the plurality of facial landmarks. The system of claim 10 further comprising:

14. The system of claim 10 , wherein at least a subset of the set of eyeblink parameters is determined using aspect ratio information calculated from a set of facial landmarks.

15. The system comprises: a system for performing simulation operations; A system for performing simulation operations to test or validate an autonomous machine application; a system for rendering graphical output; A system for performing deep learning operations, a system implemented using edge devices; a system incorporating one or more virtual machines (VMs); a system implemented at least in part in a data center; or A system implemented at least in part using cloud computing resources The system of claim 10, comprising at least one of:

16. a camera for capturing image data including a representation of a person's face over a period of time; one or more processors; A memory containing instructions; wherein the instructions, when executed by the one or more processors, cause the system to: identifying a set of facial landmarks in an image containing a representation of a human face; determining eye state information from at least a portion of the image, the eye state information indicating at least whether the person's eyes are open or closed; determining head pose information for the person based at least in part on the facial landmarks; determining a set of eyeblink parameters for the person over a recent time period based at least in part on the set of facial landmarks, the head pose information, and the eye state information; using the eye state information to determine blink frequency information for the recent time period; using the set of eyeblink parameters as input to a first temporal network to determine a first drowsiness prediction for the person; using the blink rate information as an input to a second temporal network to determine a second drowsiness prediction for the person; and generating an overall somnolence determination for the person based at least in part on the first somnolence prediction and the second somnolence prediction; and A drowsiness detection system that:

17. The instructions, when executed, cause the drowsiness detection system to: determining at least one action to take with respect to the person based at least in part on the overall somnolence determination; The drowsiness detection system of claim 16 , further comprising:

18. The instructions, when executed, cause the drowsiness detection system to: using the image to determine an identity of the person; identifying an eyeblink profile for the person, the eyeblink profile being indicative of one or more eyeblink behaviors of the person; generating at least the first drowsiness prediction further based on data about the one or more eye blink behaviors; The drowsiness detection system of claim 16 , further comprising:

19. The instructions, when executed, cause the drowsiness detection system to: performing a transformation of the set of facial landmarks based at least in part on the head pose information to adjust an orientation of the set of facial landmarks. The drowsiness detection system of claim 16 , further comprising:

20. 17. The drowsiness detection system of claim 16, wherein the first temporal network and the second temporal network are long short-term memory (LSTM) networks, and the first drowsiness prediction and the second drowsiness prediction generated by the LSTM networks correspond to Karolinska Sleepiness Scale (KSS) values.

Images