371 results about "View camera" patented technology

The invention provides an intelligent heavy truck end-to-end driving method based on focus attention and probabilistic game, and belongs to the technical field of intelligent driving. The method comprises the following steps: firstly, acquiring multi-source data such as a laser radar, a look-around camera, an IMU (Inertial Measurement Unit) and a load signal, and generating an anti-jitter space-time BEV feature through cross-dimensional position coding based on motion compensation, near-field non-uniform sampling, focus attention and time sequence fusion; then, vectorization decoding is carried out on the features to obtain obstacle vehicle movement and map element vector information. And finally, on the basis of the expert track prior space, probability game interaction is carried out through multi-round iterative prediction and planning, a self-vehicle track conforming to heavy truck variable load dynamics constraints is generated, and the self-vehicle track is output after safety verification. According to the method, the problems of heavy truck high-position sensor vibration distortion, hinge structure blind area and trajectory planning under the large-inertia variable-load working condition are effectively solved, and the driving safety and robustness are remarkably improved.

The invention belongs to the technical field of laser radar-camera panoramic segmentation, and particularly relates to a multi-mode panoramic segmentation method for multi-view space-time alignment and implicit feature interaction, and the method is executed by a multi-mode panoramic segmentation network, and comprises the steps: S1, obtaining laser radar point cloud data and multi-view camera image data in the same scene; s2, performing double-branch feature coding on the laser radar point cloud data; performing multi-scale image feature extraction on the camera image data; s3, generating false point cloud features with geometric perception capability; s4, generating semantic pixel features; s5, performing implicit fusion on the pseudo point cloud features generated in the S3 and the semantic pixel features generated in the S4 to obtain cross-modal fusion features; and S6, based on the cross-modal fusion features obtained in the S5, generating a unified panoramic segmentation result containing semantic tags and instance IDs. The method can effectively improve the robustness and precision of multi-mode panoramic segmentation in a complex urban environment.

The application relates to the technical field of blind area monitoring, and relates to a low-speed blind area monitoring method and system based on multi-view camera shooting, which comprises the following steps: registering multiple cameras in a vehicle to obtain a target vehicle and a vehicle coordinate system; confirming multiple early warning coordinates based on the vehicle coordinate system, a vehicle bird's eye view and the vehicle; confirming a candidate window based on the vehicle bird's eye view and the early warning coordinates; performing target detection on the candidate window by using a pre-constructed target detection model to obtain a detection result; if the detection result is a first result, taking a pre-constructed safety instruction as a target instruction; otherwise, confirming relative positions and relative speeds based on target bounding boxes in a second result and multiple radars in the target vehicle; calculating an expected collision time according to the relative positions and the relative speeds; confirming the target instruction based on the expected collision time; and summarizing the target instruction to obtain multiple target instructions, thereby completing low-speed blind area monitoring. The application can improve the precision of low-speed blind area monitoring.

The invention relates to the technical field of intelligent traffic and artificial intelligence, and discloses an end-to-end automatic driving method and system based on multi-modal attention fusion, and the method comprises the steps: collecting multi-modal data; generating an initial semantic text and calibrating the initial semantic text to form a calibrated semantic text; performing feature processing on the environment image and the point cloud data acquired by the multi-view camera, performing cross-modal alignment on the calibrated semantic text and the first BEV feature, and performing spatial-temporal feature modeling; generating candidate trajectories, quantifying the collision risk of each candidate trajectory and the dynamic obstacle, screening out low-risk trajectories, and guiding the low-risk trajectory optimization through a rule mask; mapping the optimized track into a control instruction; according to the method, a CLIP cross-modal alignment mechanism is utilized, text features and BEV geometric features are deeply fused, and the recognition accuracy of the key region is improved.

The invention belongs to the technical field of industrial robot motion control, and discloses an industrial robot precise path generation system based on three-dimensional visual reconstruction, comprising an environment sensing device which is composed of a multi-baseline structured light projector and a multi-view camera array and obtains point cloud data through light interference suppression and spectrum correction; the three-dimensional reconstruction processing unit generates a three-dimensional model through rigid transformation and deep learning correction; the geometric feature analysis unit performs curvature analysis and boundary recognition to form region description; the trajectory planning device adopts a weighted B spline to generate a trajectory under curvature constraint and energy optimization; the dynamic correction unit is switched between vision feedback and force feedback according to an error threshold value; the control conversion processor combines inverse kinematics and dynamics compensation and predicts deviation to output a control instruction; and the execution driving mechanism utilizes a servo motor and closed loop feedback to execute actions, so that high-precision generation and stable execution of a track under a complex working condition are realized.

The invention relates to the technical field of image shielding processing, in particular to a target graph shielding processing method and system based on variable-focus wide-view-angle cameras, and the method comprises the steps: deploying a plurality of variable-focus cameras based on seat distribution, constructing a monitoring network, and collecting an initial image; identifying desktop features according to the images, establishing a unified coordinate system and mapping each desktop; when shooting is needed, querying candidate cameras according to the coordinate system, and determining a main camera according to an imaging area and a distance score; and performing quality evaluation on the main shot image, outputting the main shot image if the main shot image reaches the standard, otherwise, triggering a dynamic re-shooting mechanism. According to the invention, through cooperation of multiple cameras and intelligent visual angle selection, the problem that a single camera is easy to block is effectively solved, and automatic and high-quality image acquisition and processing are realized.

The invention provides an automatic driving target detection method and electronic equipment, and belongs to the technical field of computer vision and automatic driving, and the method comprises the steps: obtaining multi-view image data through a vehicle-mounted multi-view camera, carrying out the preprocessing of the image data, and inputting the image data into a backbone network to extract multi-scale image features; performing channel attention weighting and space attention weighting on the multi-scale image features, and inputting the weighted features into a bidirectional feature pyramid network for bidirectional fusion to generate enhanced multi-scale fusion features; generating auxiliary target category prediction and bounding box prediction by using an auxiliary prediction head based on the multi-scale fusion features, and calculating auxiliary supervision loss according to the target category prediction and bounding box prediction so as to apply direct supervision to the backbone network; and integrating the obtained auxiliary prediction bounding box into a position query of a decoder to form a mixed query, inputting the mixed query and BEV features into the decoder for optimization, and outputting a final target detection result.

Systems and methods for navigating a host vehicle are disclosed. In one implementation, a system includes a processor configured to receive at least one image captured by a front view camera of the host vehicle; analyze the at least one image to detect a representation of a traffic sign in an environment of the host vehicle; determine a relevancy of the traffic sign to the host vehicle, wherein the relevancy of the traffic sign to the host vehicle is determined based on a trajectory of the host vehicle relative to a location of the traffic sign; and cause the host vehicle to implement at least one navigational action based on the relevancy of the traffic sign to the host vehicle.

The invention relates to a deep feature fusion-based fabricated component surface defect multi-task detection method. The method comprises the following steps of: constructing a multi-view camera array and an illumination feedback regulation and control module at a data acquisition end; on the algorithm level, feature cross-layer propagation and information compensation are realized through an improved lightweight convolutional network and a multi-scale residual diffusion module; a dynamic feature fusion module is introduced, and a self-adaptive fusion weight is generated based on channel statistical features, so that feature sharing and differential expression are realized among different tasks; meanwhile, a defect perception attention mechanism and cross-task consistency constraint are adopted, and the problem that semantic space distribution is inconsistent in the multi-task detection process is solved; in the detection post-processing stage, a three-dimensional quantitative evaluation system based on the geometric dimension, the texture roughness and the depth volume is constructed, and unified grade evaluation of the surface defects of the component is achieved through the multi-feature fusion quality index. The problems of low detection efficiency, unstable precision, difficulty in collaborative recognition of multiple types of defects and the like in existing component delivery detection are solved.

An interactive gaming system comprising a multiplicity of game access points used by guests to engage with an on-going destination game. Game access points comprise traditional destination activities as well as additional game parts. Gaming guests carry one or more electronic means and are tracked throughout the destination locations and activities. At least one electronic means comprises a visual display for receiving requested directions or a system summons to a game access point. The electronic means comprises a direct view lens system such that the guest looks directly through the electronic means to a destination map or destination scene, where directions are overlaid onto the direct view and registered to the map or scene. The mobile device further comprises a front view camera for capturing images of the guest's direct view of a destination game part, where the image is used as actionable game information.

The invention discloses a transformer substation video fusion monitoring method and system based on 3DGS, and relates to the technical field of three-dimensional modeling. The method comprises the following steps: acquiring a laser radar point cloud and a multi-view camera image of a transformer substation, and constructing a three-dimensional Gaussian splashing model; establishing a mapping relation library based on P source view angle and Q target view angle transformer substation 3DGS samples; based on the library, using a picture splicing boundary confidence degree adjusting pointer and a space coordinate mapping deviation correction pointer to carry out fusion training; and when a user clicks a target device in the model, calling an improved line-of-sight cone analysis algorithm to dynamically screen an optimal camera, receiving a video stream of the optimal camera, and determining a device state recognition result. The technical problem of poor accuracy of image splicing and equipment positioning caused by multi-view and view range limitation in substation equipment monitoring is solved, and the technical effect of improving the monitoring precision and efficiency of substation equipment by dynamically screening the optimal camera and equipment state recognition of space coordinates is achieved.

The invention relates to the technical field of rubber detection, in particular to a rubber part quality detection device and a detection method thereof. The device comprises a detection index plate, a support and two lifting devices are sequentially arranged on the periphery of the detection index plate, a laser diameter measuring sensor is arranged on the support, and the laser diameter measuring sensor is arranged above a vibration disc feeder in a suspended mode and used for measuring the diameter of a passing rubber part. A top view camera and a side view camera are connected to the lifting device, an annular lamp is fixedly arranged on the periphery of a lens of the top view camera, a supporting rod is fixedly connected to the lifting device corresponding to the side view camera, and a parallel backlight lamp is arranged at the end of the supporting rod; light shields are arranged on the peripheries of the annular lamp and the parallel backlight lamp, baffles are movably arranged in the light shields, and driving assemblies are connected between the baffles and the light shields; the driving assembly adjusts the rotation angle of the laser diameter measuring sensor according to detection data of the laser diameter measuring sensor, drives the baffle to move in the light shield, and adjusts the local shielding range of the baffle to the light emitting area of the corresponding light source.

The invention discloses a multi-view camera splicing monitoring system based on a sky target, and relates to the technical field of visual intelligence. The multi-view camera splicing monitoring system based on the sky target comprises the following steps: a multi-view data acquisition module used for acquiring multiple types of data sets and carrying out time alignment and dimensionless normalization; the target detection and luminosity reference construction module is used for carrying out sky target detection and standard color block analysis and constructing a multi-view luminosity reference library; the deviation estimation and mapping construction module is used for constructing a cross-camera global luminosity mapping model; the adaptive control and real-time correction module is used for regional luminosity correction and illumination change detection; and the panoramic stitching and evaluation feedback module is used for constructing a joint area luminosity consistency evaluation index. According to the method, the multi-view splicing brightness color consistency and the target detection robustness under complex illumination are effectively improved, and the problems of seam brightness gradient and difficult color cast elimination caused by independent automatic exposure and white balance are solved.

Rear-view image(s) of a region of a surrounding environment that is behind a vehicle, is / are captured, by utilising rear-view camera(s). An image to be displayed via the heads-up display, is generated, wherein when generating the image, processor(s) is / are configured to generate an image segment of the image by utilising the rear-view image(s) of said region of the surrounding environment. The image is displayed via the heads-up display for producing a synthetic light field augmenting a real-world light field incoming via a windshield of the vehicle.

The application relates to the technical field of intelligent auxiliary reminding, in particular to an auxiliary reminding method based on a smart wearable device, which comprises the following steps: collecting behavior data of the elderly through a multi-view camera, an infrared sensor and a pressure sensing device, constructing a four-dimensional characteristic behavior sequence graph, and using a deep learning model with a multi-modal fusion and attention enhancement mechanism to analyze behavior semantics and generate graded reminding instructions. The application can accurately monitor the daily behavior of the elderly, dynamically adjust the reminding strategy, trigger an emergency alarm in an abnormal situation and record intervention logs, and effectively improve the life safety of the elderly and the monitoring efficiency.

1. The name of the design product: automobile traveling data recorder (YKC-AEB101). 2. The use of the design product: the design product is used for commercial vehicle auxiliary driving control, built-in double hardware system, and product functions integrate millimeter wave radar, ADAS, front view camera, DMS, BSD, AEBS, alarm display, display screen, brake actuator, etc. 3. The design points of the design product: in shape. 4. The picture or photo that best indicates the design points: perspective view 1.

The invention relates to the field of camera calibration, in particular to a calibration method, device and equipment for a yaw angle of a vehicle foresight camera and a medium. The method comprises the following steps: acquiring a driving direction information sequence of a vehicle in a straight driving state; according to the driving direction information sequence, obtaining a reference driving direction of the vehicle driving along a straight line; determining the actual driving direction of the vehicle at the target moment based on the driving direction information at the target moment in the driving direction information sequence; determining a compensation yaw angle according to the deviation between the actual driving direction and the reference driving direction; and on the basis of the compensation yaw angle, calibrating an estimated value of the external parameter yaw angle of the foresight camera at the target moment, and obtaining a calibration value of the external parameter yaw angle of the foresight camera.

The invention relates to the technical field of image processing, in particular to a bogie part image recognition and positioning method and system based on deep learning, and the method comprises the steps: constructing a bogie part and bolt image data set, and carrying out the preprocessing and marking; a recognition positioning model (introducing an ECA attention mechanism, optimizing a BiFPN feature fusion structure and designing a special bolt detection head) is constructed based on an improved YOLOv8 algorithm, after the model is trained and optimized, images are collected through a multi-view camera and input into the model to obtain two-dimensional coordinates, the two-dimensional coordinates are converted into three-dimensional space coordinates in combination with camera calibration and binocular vision, and the three-dimensional space coordinates are obtained. And after post-processing optimization, transmitting to a robot control system to guide full-automatic decomposition. According to the method, the recognition accuracy and positioning precision of the bogie parts and the bolts in the complex environment are improved, the bolt positioning error is within + / -0.5 mm, the recognition accuracy is larger than or equal to 98%, and support is provided for intelligent bogie maintenance.

A method of compensating for non-deterministic errors in a robotic cell including a robot arm with an end of arm tool, one or more static cameras on the robotic cell, and one or more tip-view cameras on the robot arm, the method comprising using the one or more static cameras for targeting movement during a first phase of movement to a first position, and using the one or more tip-view cameras for targeting the movement during a second phase or subsequent phases, images from the tip-view cameras used to calculate an offset defining a difference between the first position and an intended position.

A computer that includes a processor and a memory, the memory including instructions executable by the processor to determine top key points from one of an aerial feature map or one or more ground feature maps and project the top key points as corresponding lines on the other of the aerial feature map or the one or more ground feature maps. The memory includes instructions to determine a depth estimate for the top key points on the corresponding lines. A high-definition estimated three degree-of-freedom pose of a ground view camera is determined in global coordinates by iteratively determining geometric correspondence between the top key points and the corresponding lines until a global loss function is less than a user determined threshold.

The application discloses a parking effect evaluation method, an electronic device and a storage medium. The parking effect evaluation method comprises the following steps: acquiring first image data collected by a vehicle-mounted surround-view camera; performing image processing on the first image data to obtain a target bird's-eye view; performing parking space detection processing on the target bird's-eye view to determine parking space corner point coordinates; and evaluating the parking effect according to the parking space corner point coordinates and vehicle size parameters. In this way, the first image data is collected by the vehicle-mounted surround-view camera, the target bird's-eye view is obtained through image processing, the parking space corner point coordinates are determined through parking space detection, and the parking effect evaluation is completed in combination with the parking space corner point coordinates and the vehicle size parameters, thereby solving the problems that the automatic parking effect evaluation depends on preset scenes and other additional auxiliary conditions, the adaptation range is limited, and the evaluation data is difficult to obtain, and effectively improving the accuracy of the evaluation result and the ability to adapt to complex and variable actual parking scenes.

A vehicle includes: camera monitoring apparatus including a side-view camera having a field of view facing a rear lateral side of the vehicle and a side-view display configured to display a side-view image captured by the side-view camera; and a driver assistance system including a corner radar having a field of view facing a side of the vehicle, and configured to process radar data of the corner radar. The driver assistance system is configured to provide the camera monitoring apparatus with a virtual side-view image based on processing of the radar data of the corner radar when a failure of the side-view camera is detected. The camera monitoring apparatus is configured to display the virtual side-view image of the driver assistance system on the side-view display based on the failure of the side-view camera.

The invention discloses a vehicle charging port cover control method and device, equipment and a medium. The method comprises the steps that first rainfall data detected by a rainfall sensor of a vehicle and second rainfall data shot by a front-view camera of the vehicle are obtained in real time; determining a current weather type according to the first rainfall data and the second rainfall data; if the weather type indicates that the current weather is rain and snow weather, the opening state of a charging port cover of the vehicle and the insertion state of a charging gun of the vehicle are obtained; and if the charging port cover is in the open state and the charging gun is in the non-inserted state, controlling the charging port cover to be closed. The vehicle can automatically close the opened charging port cover in time when a user is not near the vehicle and cannot close the port cover in rainy and snowy weather.

The application discloses a ranging method, device and system based on camera and radar fusion and a medium. In order to solve the problem that when a shovel remotely controls operation through a remote control console outside the visual range, the remote control end often does not have a good operation visual angle, the perception of the excavated material depth information is not intuitive, the position perception of the excavated material area is not intuitive, and the shovel cannot be accurately lowered, a method for fusing a front-view camera of a shovel and laser radar point clouds is provided. Through the new radar and visual fusion technology, the method can realize video monitoring of the material area in front of the shovel, and video pictures are superimposed to display point clouds of the excavated material, target material point clouds and operation information.

The application discloses a vehicle seat occupancy detection method and system and a storage medium, and relates to the technical field of vehicle intelligent detection. The method comprises the following steps: in response to an opening operation of a target vehicle door, camera sensing data in a preset time period is acquired, wherein the camera sensing data is used to represent vital sign object information sensed by a surround-view camera of a target vehicle within a preset range of the target vehicle door; and based on the camera sensing data and seat pressure sensor data, the occupancy of a vehicle seat corresponding to the target vehicle door is determined, wherein the seat pressure sensor data is used to represent force information of the vehicle seat sensed by a seat pressure sensor of the target vehicle. The application solves the technical problem that the accuracy of vehicle seat occupancy detection in the related art is low, and thus the vehicle safety reminder function cannot be effectively triggered to ensure passenger safety.

This invention discloses a color balancing method, apparatus, electronic device, and storage medium. The method includes: acquiring four video images captured by a surround-view camera; performing distortion correction on the four video images to obtain four target video images; wherein the four video images are ultra-wide-angle images of the target object in four directions (front, back, left, and right); performing inverse perspective transformation on the four target video images to generate four top-view images; and performing color balancing on any one top-view image based on curve fitting to obtain the target top-view. This invention achieves color balancing through curve fitting, solving the problems of poor adaptability and low robustness of existing methods, and improving the adaptability and robustness of color balancing methods.

The invention relates to a light finding clamping device for optical lens assembly and a method thereof, and belongs to the technical field of optical lens clamping, and the device comprises a positioning host which comprises a base, an upper view camera and a multi-partition annular light source; wherein the base is provided with an upward through hole, and a frosted glass plate for defining a lens placing plane is mounted in the through hole; the upper view camera is fixed in the base and located under the frosted glass plate, and the optical axis of the upper view camera is perpendicular to the placement plane; the multi-partition annular light source is coaxially mounted on the outer ring of a lens of the upward-looking camera; and the gantry type mechanical arm is erected above the positioning main machine and comprises a flexible clamping mechanism, the clamping action can directly aim at the optical center of the lens for alignment, and the coaxiality problem is solved.

A detection method and apparatus, and a vehicle. The method comprises: acquiring a first group of images collected by a first camera apparatus of a vehicle, wherein the first group of images comprises at least one image frame; when the first group of images does not comprise pixels corresponding to a first object, controlling a second camera apparatus of the vehicle to acquire a second group of images, wherein the second group of images comprises at least one image frame; and on the basis of the second group of images, determining information of the first object. The technical solution can be applied to the field of intelligent driving of intelligent vehicles such as electric vehicles and new energy vehicles. When a front-view camera apparatus of a vehicle cannot detect road information, a camera apparatus with a variable yaw angle is controlled to acquire images in the orientation of the road information, such that relevant road information can be detected in a timely manner, thereby helping to ensure travel safety and traffic efficiency.

The application relates to a method and system for realizing bird's-eye view semantic segmentation based on implicit set prediction, which comprises the following steps: preprocessing labeled data in a high-precision map, converting a semantic layer of the high-precision map into a labeled polygon layer; extracting semantic features in the labeled polygon layer, performing prediction head analysis on the semantic features, realizing splicing and fusion of multi-view cameras, and finally obtaining a polygon segmentation frame through implicit set prediction; and converting an image plane into a perspective of a bird's-eye view, and completing implicit prediction based on a transformer. The application can make a network simultaneously perceive dynamic objects and static roads, and reduce the calculation amount of the perception network; and can be applied in the field of automatic driving.

A vehicular vision system includes an electronic control unit (ECU) having a plurality of coaxial connectors. Individual vehicular cameras of a plurality of vehicular cameras are connected to respective coaxial connectors of the ECU via respective coaxial cables. Electrical power for a vehicular forward-viewing camera is carried from the ECU to the vehicular forward-viewing camera via a first coaxial cable connected to a first coaxial connector of the ECU. Image data captured by the vehicular forward-viewing camera is carried to the ECU via the first coaxial cable and is converted at a first serializer of the camera to first serialized image data and is de-serialized at a first de-serializer of the ECU to form first de-serialized image data. Responsive at least in part to processing of first de-serialized image data, an object is detected. Processing of first de-serialized image data at the ECU includes structure from motion (SFM) processing.