Image-based vehicle cabin sensing and vehicle device operation

Abstract

The invention relates to image-based vehicle cabin sensing and vehicle equipment operation. This includes a method of facilitating operation of a vehicle, the method comprising the steps of: detecting one or more seat components of at least one seat in a vehicle cabin from one or more images of at least a portion of the vehicle cabin, and detecting a seat component property of each detected seat component; and estimating one or more seat parameters of the at least one seat based on the detected one or more seat component attributes, the one or more seat parameters of the seat indicating a spatial configuration of the seat in the vehicle cabin.

Description

Technical Field

[0001] The present disclosure relates generally to safety, comfort, and / or personalization improvements for vehicles, and particularly to methods, devices, systems, and computer program products for facilitating vehicle operations, such as sensing a vehicle cabin and / or operating vehicle devices. Background Art

[0002] Modern vehicles such as cars, buses, trucks, etc. are often equipped with one or more onboard cameras, which are cables that capture images of the vehicle interior (such as the vehicle cabin). These images can be used for various purposes of vehicle control. For example, these images can be used, potentially together with input from sensors disposed in the vehicle interior (e.g., the vehicle cabin) to control various elements of vehicle equipment (e.g., air bags).

[0003] The present disclosure aims to provide improvements with respect to using images from one or more vehicle-mounted cameras. Summary of the invention

[0004] According to one aspect of the present disclosure, a method is provided, which may be considered or referred to as a method for facilitating vehicle operation or a method for facilitating vehicle operation and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method. The method comprises detecting one or more seat components of at least one seat in the vehicle cabin from one or more images of at least a portion of the vehicle cabin, and detecting seat component attributes of each detected seat component, and estimating one or more seat parameters of at least one seat based on the detected one or more seat component attributes (of the corresponding seat), the one or more seat parameters of the seat indicating a spatial configuration of the seat in the vehicle cabin.

[0005] In an embodiment, the seat component attributes include a detected bounding box of the seat component that surrounds the seat component in one or more images and / or a specific point or location of the seat component detected in one or more images.

[0006] In an embodiment, geometric modeling is used to estimate one or more seat parameters. In other embodiments, when geometric modeling is used, estimating one or more seat parameters includes determining geometric features of at least a portion of the detected one or more seat component attributes and / or a relationship between one or more groups of at least two detected seat component attributes, and calculating the one or more seat parameters based on the determined geometric features. In this regard, the geometric features may include one or more of the position of the seat component attribute, the size of the seat component attribute, the shape of the seat component attribute, and the relative two-dimensional, horizontal and / or vertical position between two or more seat component attributes. Additionally or alternatively, calculating the one or more seat parameters may utilize a linear least squares method or a Levenberg-Marquart algorithm.

[0007] In an embodiment, a neural network is used to estimate one or more seat parameters. In further embodiments, when a neural network is used, estimating one or more seat parameters includes inputting the detected one or more seat component attributes into the neural network and receiving the one or more seat parameters as outputs of the neural network. In this regard, the neural network can be a trained neural network that can be trained using training data defining seat component attributes as inputs and seat parameters as outputs and / or ground truth data defining seat parameters for a predetermined spatial configuration of seats in a vehicle cabin.

[0008] In an embodiment, a lookup table is used to estimate one or more seat parameters. In further embodiments, when a lookup table is used, estimating one or more seat parameters includes deriving one or more seat parameters from the lookup table using one or more detected seat component attributes as parameters. In this regard, the lookup table may be a calibrated / preconfigured lookup table that may be calibrated / preconfigured using various constellations of seat parameters (which may indicate one or more seat component attributes, such as specific points or locations of seat components detected in one or more images).

[0009] In an embodiment, the seat components of the seat include one or more of a seat cushion, a seat back, a seat headrest and a seat belt buckle, and / or the seat parameters of the seat include one or more of a seat track position, a seat recline angle and a seat headrest position.

[0010] In an embodiment, one or more images are captured by an image capture device, which is arranged in the vehicle cabin, its field of view includes one or more seats in the vehicle cabin and / or is configured to capture color and / or infrared images.

[0011] According to one aspect of the present disclosure, a method is provided, which may be considered or referred to as a method for facilitating vehicle operation or a method for facilitating vehicle operation and / or a method for operating vehicle equipment or an operation method (of vehicle equipment). The method comprises the following steps: estimating one or more seat parameters of a seat in a vehicle cabin where a person sits; detecting one or more body parts of the person from one or more images of at least a portion of the vehicle cabin, and detecting body part attributes of each body part; and controlling the vehicle equipment based on the estimated one or more seat parameters and the detected one or more body part attributes of the person.

[0012] In an embodiment, controlling the vehicle equipment includes adjusting one or more of a setting or configuration of the vehicle equipment and causing a predetermined operation of the vehicle equipment. In this regard, the vehicle equipment may include one or more of a seat, a seat component of the seat, a controller, an airbag, and an acoustic and / or visual output device. In addition, when the vehicle equipment is at least one seat component of the seat, controlling the vehicle equipment may include adjusting the at least one seat component so that the seat parameters corresponding to the at least one seat component are adapted to one or more detected body part attributes of the person.

[0013] In an embodiment, the body part attributes include one or more of: a bounding box of a detected body part, the bounding box of the body part surrounding the body part in one or more images; one or more key points of the detected body part, the one or more key points of the body part defining feature points of the body part in one or more images; and an orientation of the detected body part, the orientation of the body part defining a spatial orientation of the body part in a vehicle cabin.

[0014] In an embodiment, the body part of the person includes one or more of the person's head and torso.

[0015] In an embodiment, the one or more images are captured by an image capture device, the vehicle camera being arranged in the vehicle cabin and having a field of view including the seats and / or being configured to capture color and / or infrared images.

[0016] In an embodiment, by executing the above-mentioned method to estimate seat parameters, the method may be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitation method and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method, the method including any one or more embodiments thereof.

[0017] According to one aspect of the present disclosure, there is provided a device configured to perform the aforementioned method, which may be viewed as or referred to as a method for facilitating vehicle operation or a method for facilitating vehicle operation and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method, including any one or more embodiments thereof.

[0018] In an embodiment, the apparatus comprises one or more means for performing the functions / steps of the method, such as means for detecting one or more seat components and detecting seat component properties of each detected seat component; and means for estimating one or more seat parameters of at least one seat based on the detected one or more seat component properties.

[0019] In an embodiment, the apparatus comprises at least one processor and program code or circuitry configured to perform the functions / steps of the method, such as detecting one or more seat components, and detecting seat component properties of each detected seat component, and estimating one or more seat parameters of at least one seat based on the detected one or more seat component properties.

[0020] According to one aspect of the present disclosure, a device configured to perform the aforementioned method is provided, which method can be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitation method and / or a method for operating vehicle equipment or an operation method (of vehicle equipment), including any one or more embodiments thereof.

[0021] In an embodiment, the apparatus includes one or more devices for performing the functions / steps of the method, such as a device for estimating one or more seat parameters of a seat in which a person sits in a vehicle cabin, a device for detecting one or more body parts of a person and detecting body part attributes of each body part; and a device for controlling vehicle equipment.

[0022] In an embodiment, the device includes at least one processor and program code or circuit, which is configured to perform the functions / steps of the method, such as estimating one or more seat parameters of a seat in which a person sits in a vehicle cabin, detecting one or more body parts of the person, and detecting body part attributes of each body part, and controlling vehicle equipment.

[0023] According to one aspect of the present disclosure, a system is provided. The system includes an image capture device configured to capture one or more images of at least a portion of a vehicle cabin, at least a portion of the vehicle cabin including at least one seat in the vehicle cabin. The system also includes a device configured to perform the aforementioned method, which can be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method, including any one or more embodiments thereof, and / or a device configured to perform the aforementioned method, which can be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method and / or a method for operating a vehicle device or a (vehicle device) operating method, including any one or more embodiments thereof.

[0024] According to one aspect of the present disclosure, a computer program product is provided, which, when executed on a computer, causes the computer to execute the aforementioned method, which may be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method, and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method, including any one or more embodiments thereof, and which may be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method, and / or a method for operating vehicle equipment or an operating method (of vehicle equipment), including any one or more embodiments thereof.

[0025] According to one aspect of the present disclosure, a method is provided, which combines (i.e. includes) the operations / functions of the aforementioned methods, and the method can be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method and / or a method for sensing a vehicle cabin or a (vehicle cabin) sensing method, including any one or more embodiments thereof, and the method can be regarded as or referred to as a method for facilitating vehicle operation or a vehicle operation facilitating method and / or a method for operating a vehicle device or a (vehicle device) operating method, including any one or more embodiments thereof. In addition, corresponding methods, corresponding systems, and corresponding computer program products are provided.

[0026] These and other objects, embodiments and advantages will become apparent to those skilled in the art from the following detailed description of exemplary embodiments with reference to the accompanying drawings, the present disclosure being not limited to any particular embodiment. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The foregoing and other objects, features, and advantages of the present subject matter will become apparent from the following description of exemplary embodiments with reference to the accompanying drawings, wherein like reference numerals are used to designate like elements.

[0028] Figure 1 is a flow chart illustrating a method of sensing a vehicle cabin.

[0029] FIG. 2A to FIG. 2D is a schematic diagram showing exemplary detection results including an example of a seat component and an example of a seat component attribute, and Figure 2E is a schematic diagram showing a mesh cube representing a lookup table for estimating one or more seat parameters.

[0030] Figure 3 is a flow chart illustrating a method of operating a vehicle device.

[0031] Figure 4 is a diagram illustrating exemplary detection results including examples of body parts and bounding boxes as examples of attributes of the body parts.

[0032] Figure 5 is a schematic diagram illustrating exemplary detection results including examples of body parts and key points as examples of attributes of the body parts.

[0033] Figure 6 is a schematic diagram illustrating exemplary detection results including examples of body parts and orientations as examples of attributes of the body parts.

[0034] Figure 7 is a schematic diagram showing an exemplary interior of a vehicle with possible camera locations.

[0035] Figure 8 is a schematic diagram illustrating an exemplary representation of a computing system. DETAILED DESCRIPTION

[0036] Modern (or smart) vehicles (e.g., cars, buses, trucks, etc.) are on the road to significantly improving the safety, comfort, and / or personalization of passengers. Such smart vehicles may be equipped with one or more onboard cameras and are capable of capturing images of the vehicle interior (e.g., the vehicle cabin). These images may then be used in conjunction with other sensors (e.g., mechanical sensors or Hall Effect sensors) disposed at the vehicle seats to understand the conditions within the vehicle.

[0037] For different safety, comfort and / or personalization related tasks, certain seat parameters are relevant and therefore need to be detected. These seat parameters include, for example, seat track position, seat recline angle (i.e. seat back angle) and seat headrest position. In order to detect these seat parameters, sensors, such as mechanical sensors or Hall effect sensors, are currently required to be arranged at the vehicle seat.

[0038] Therefore, in order to correctly detect seat parameters relevant to different safety, comfort and / or personalization related tasks, seat equipment with sensors is required, which leads to increased costs and cabling / wiring work, as well as problems and less flexibility for retrofitting / upgrading the corresponding functionalities.

[0039] Although it would be desirable to be able to detect such seat parameters in a sufficiently reliable and / or accurate manner without requiring any seat-equipped sensors, there is currently no solution to this problem.Therefore, there is room for improvement in terms of safety, comfort and / or personalization improvements in vehicles.

[0040] The present disclosure relates generally to safety, comfort, and / or personalization improvements for vehicles, and particularly to methods, devices, systems, and computer program products for sensing a vehicle cabin and / or operating vehicle equipment.

[0041] In this regard, certain seat parameters need to be detected, such as seat track position, seat recline angle (ie, seat back angle), and seat headrest position, which typically requires a seat device having sensors (eg, mechanical sensors or Hall effect sensors).

[0042] The present disclosure proposes a solution for vehicle cabin sensing and vehicle equipment operation with reference to seat parameters without the need for any seat-equipped sensors. More specifically, the present disclosure provides a technique for image-based vehicle cabin sensing and vehicle equipment operation, in which the relevant seat parameters are determined optically. Even if the relevant seat parameters cannot be directly measured optically, the technique of the present disclosure is able to estimate the seat parameters in vehicle cabin sensing based on one or more images of at least a portion of the vehicle cabin, and thereby enable the use of the seat parameters in vehicle equipment operation, while avoiding or at least reducing ambiguity in the detection of the seat parameters.

[0043] As an illustrative but non-limiting example, the seat track position and / or seat recline angle may be optically estimated in vehicle cabin sensing as a seat parameter of the driver's seat, and the driver's seat (i.e., its positioning) (as an example of a vehicle device) may be controlled based on the seat track position and / or seat recline angle so estimated in vehicle device operation. For example, the driver's seat may be (controlled to be) operated such that the seat track position and / or seat recline angle are changed, such as the seat being driven toward the steering wheel and / or the seat backrest being driven to a vertical position, for enabling the driver to take over control of the vehicle when needed, such as in a situation where a Level 3 autonomous driving mode is active but needs to be deactivated.

[0044] It should be noted that the present disclosure is not limited to the estimation of the seat parameters of the driver's seat, but can be applied to any seat in the vehicle cabin. For example, the seat parameters of one or more front seats (e.g., driver's seat and / or passenger seat) can be estimated.

[0045] Figure 1is a flow chart illustrating a method according to at least one embodiment, which method may be viewed or referred to as a method of facilitating vehicle operation or a vehicle operation facilitation method and / or a method of sensing a vehicle cabin or a (vehicle cabin) sensing method.

[0046] like Figure 1 As shown, the method is based on one or more images of at least a portion of the vehicle cabin, and thus may include inputting or receiving an image (block 11). The one or more images may be captured by an image capture device (e.g., a camera) that is arranged in the vehicle cabin, has a field of view that includes one or more (or all) seats in the vehicle cabin (and / or passengers sitting on the one or more seats), and / or is configured to capture color images and / or infrared images. For example, an RGB-IR camera that can capture RGB (red, green, blue) colors as well as the IR (infrared) spectrum may be used.

[0047] Modern vehicles, in particular cars, may be equipped with one or more cameras. These cameras may include one or more of a color or black / white camera, an infrared camera, a depth camera, and a thermal camera, which may provide images of the vehicle cabin to internal components of the vehicle, such as devices and / or computing systems. Images may be captured in response to a triggering event such as opening a door, interacting with the vehicle's multimedia system, using voice commands, etc., or at defined periodic or aperiodic time intervals. The captured images may then be used by multiple internal systems for different tasks, such as safety, comfort, and / or personalization related tasks.

[0048] The image of the vehicle cabin may include a complete image showing the entire vehicle cabin (or interior), or may depict only a partial area of ​​the vehicle cabin (or interior), such as one front seat, one or more rear seats, etc. If the camera captures the entire vehicle cabin (or interior), then Figure 1 The images used in the illustrated method may be pre-processed image crops that have been extracted from one or more original images of one or more regions within the vehicle cabin (or interior).

[0049] Figure 1 The method shown includes detecting one or more seat components of at least one seat in a vehicle cabin from one or more images, and detecting seat component attributes of each detected seat component (box 12), and estimating one or more seat parameters of at least one seat based on the detected one or more seat component attributes of the corresponding seat (box 13).

[0050] In the method, one or more seat parameters may be estimated in any feasible manner. Figure 1As shown, the seat parameters can be estimated using geometric modeling (box 13-1) or using a neural network (box 13-2) or using a lookup table (box 13-3). When geometric modeling is used, the seat parameter estimation may include determining a geometric feature of at least a portion of one or more detected seat component attributes and / or a relationship between one or more groups of at least two detected seat component attributes (box 13-1-1), and calculating one or more seat parameters based on the determined geometric feature (box 13-1-2). When a neural network is used, the seat parameter estimation may include inputting the detected one or more seat component attributes into the neural network (box 13-2-1), and receiving one or more seat parameters as outputs of the neural network (box 13-2-2). When a lookup table is used, the seat parameter estimation may include using the detected one or more seat component attributes as parameters to obtain one or more seat parameters from the lookup table (box 13-3-1).

[0051] It should be noted that at least methods / estimations based on geometric modeling and methods / estimations based on neural networks can also be combined. For example, the determined geometric features can be used (in addition or only) as input to a (correspondingly trained) neural network. For example, the (correspondingly trained) neural network can be used to determine the geometric features and / or the neural network (correspondingly trained) can be used to calculate seat parameters based on geometric features determined in any way.

[0052] The one or more seat parameters of the seat indicate the spatial configuration of the corresponding seat and may include, for example, one or more of a seat track position, a seat recline angle (i.e., a seat back angle), and a seat headrest position. The one or more seat components of the corresponding seat may include, for example, one or more of a seat cushion, a seat back, a seat headrest, and a seat belt buckle. The one or more seat component attributes may include, for example, a bounding box of a detected seat component, wherein the bounding box of the seat component surrounds the seat component in one or more images.

[0053] Despite Figure 1 Not shown, the method may also include estimating one or more seat parameters of a seat (in one or more seats) in which a person is seated in a vehicle cabin, detecting one or more body parts of the person from the one or more images, and detecting body part attributes of each body part, and controlling vehicle equipment based on the estimated one or more seat parameters and the detected one or more body part attributes of the person, wherein the control may include one or more of the following: adjusting settings or configurations of the vehicle equipment, and causing predetermined operations of the vehicle equipment.

[0054] The vehicle equipment may include, for example, one or more of a seat, a seat component of a seat, a controller, an airbag, and an acoustic and / or visual output device. The body part of a person may include, for example, one or more of a head and a torso of the person. The one or more body part attributes may include, for example, one or more of the following: a bounding box of a detected body part, wherein the bounding box of the body part surrounds the body part in one or more images; one or more key points of the detected body part, wherein the one or more key points of the body part define feature points of the body part in one or more images; and an orientation of the detected body part, wherein the orientation of the body part defines a spatial orientation of the body part in the vehicle cabin.

[0055] For further details on optional additional operations / functionality of estimation, detection and control as described above, reference is made to the description of the method, which may be viewed or referred to as a method of facilitating vehicle operation or a vehicle operation facilitation method and / or a method of operating a vehicle device or a (vehicle device) operation method, such as Figure 3 shown.

[0056] It should be noted that although Figure 1 is a flowchart showing a method, but it can also be regarded as a block diagram showing a corresponding device or its internal structure / components. That is, it is configured to perform Figure 1 The apparatus of the method shown may include means, modules, components, circuits, etc. for performing any one or more of the operations / functions shown in blocks 11 , 12 , and 13 .

[0057] In the following, examples and / or details of the above-mentioned operations / functionalities of detecting seat components and related seat component properties are described.

[0058] like FIG. 2A to FIG. 2D As shown, specific components of a seat can be detected in one or more images (e.g., in an image frame). For any such detected seat component, at least one seat component attribute is detected (as a basis for subsequent seat parameter estimation). That is, any seat component attribute can be detected optically or in other words in an image-based manner and is relevant or effective for supporting the estimation of one or more seat parameters. Therefore, the seat component attribute can be regarded as or referred to as a seat parameter or a proxy parameter for seat parameter estimation.

[0059] The detection of seat parts and seat part attributes can be implemented in several ways, such as any type of object detection. For example, such optical or image-based (object) detection can be performed or assisted by various systems, concepts or methods, such as convolutional neural networks, such as region-based convolutional neural networks (R-CNN), fast R-CNN, YOLO (You Only Look Once), etc.

[0060] Figure 2A Exemplary detection results are shown, in which, with respect to the front seat shown on the right-hand side of the image, a bounding box is detected around the seat headrest (as an example of a seat component) represented by 21A (as an example of a seat component) (as an example of a seat component attribute), a bounding box is detected around the seat back (as an example of a seat component) represented by 22A (as an example of a seat component) (as an example of a seat component attribute), and a bounding box is detected around the seat cushion (as an example of a seat component) represented by 23A (as an example of a seat component attribute).

[0061] Figure 2B An exemplary detection result is shown, in which the front seats shown on the right hand side of the image are detected as follows Figure 2A Similar bounding boxes are shown, which are indicated by 21B, 22B and 23B.

[0062] Figure 2C An exemplary detection result is shown, in which the front seats shown on the right hand side of the image are detected as follows Figure 2A Similar bounding boxes are shown, which are indicated by 21C, 22C and 23C. In addition, a bounding box (as an example of a seat component attribute) is detected for (ie, surrounding) a seat belt buckle (as an example of a seat component) indicated by 24C.

[0063] Figure 2D An exemplary detection result is shown, in which the front seats shown on the right hand side of the image are detected as follows Figure 2A Similar bounding boxes are shown, which are indicated by 21D, 22D and 23D.

[0064] It should be noted that one or more seat component attributes may be detected in each image or image frame, such as any one or more of the exemplarily shown bounding boxes for a seat headrest (see 21A, 21B, 21C, 21D), a seat back (see 22A, 22B, 22C, 22D), a seat cushion (see 23A, 23B, 23C, 23D), and a seat belt buckle (see 24C).

[0065] although FIG. 2A to FIG. 2D Detection results are shown exemplarily for the front seats shown on the right-hand side of the image, but similar detection results can be obtained for any seat in the image, including the front seats on the left-hand side of the image and one or more of any rear seats, as long as the seats are depicted in the image. FIG. 2A to FIG. 2D Detection results for an empty vehicle cabin are shown in , but detection results for one or more seats being occupied by people can also be obtained.

[0066] Hereinafter, examples and / or details of the above-described operation / function of estimating seat component parameters are described.

[0067] As mentioned above, seat component attributes (e.g. FIG. 2A to FIG. 2D Any of the bounding boxes shown in , which can be viewed or referred to as proxy parameters) are used as the basis for seat parameter estimation. This can be achieved in several ways.

[0068] By way of example, geometric modeling may be used to estimate seat parameters.

[0069] First, a geometric feature of a relationship between at least a portion of one or more detected seat component attributes and / or one or more groups of at least two detected seat component attributes is determined. In this regard, the seat component attributes may be encoded as a plurality of features.

[0070] For example, referring to the above bounding box, these features may include one or more of the following: the position of the detected bounding box (e.g., the center point in 2D image coordinates, such as (x, y), FIG. 2A to FIG. 2D ); the size and / or shape of the detected bounding boxes (e.g., width and height); the relative horizontal and / or vertical 2D positions between two or more detected bounding boxes (e.g., the relative positions of the center points of the detected bounding boxes in the horizontal and / or vertical directions, the relative positions of the upper and / or lower limits of the detected bounding boxes, such as the relative vertical positions of the upper limit of the bounding box of the seat back and the lower limit of the bounding box of the seat headrest). In general, any absolute and / or relative size or measure can be determined / encoded, which is effective in helping to estimate the expected seat parameters.

[0071] It should be noted that camera properties such as field of view, position (such as relative position with respect to the seat in question), angle and / or distance may also be considered / adopted in the geometric feature determination / encoding.

[0072] Secondly, one or more seat parameters are calculated based on the determined / encoded geometrical features. In this regard, a linear least squares method or a Levenberg-Marquart (LM) algorithm may be used.

[0073] For example, given the actual (3D) positions (or spatial configurations) of the seat headrest, seat back, and seat belt buckle, an equation may be defined for the positions of the seat headrest and seat belt buckle in 2D image coordinates, which depends on the seat track position and seat recline angle (as examples of seat parameters to be estimated). The equation may then be inverted to produce the seat track position and seat recline angle from the 2D (image) positions of the seat headrest and seat belt buckle. Furthermore, for the 2D (image) positions of the seat headrest and seat belt buckle, a system of equations may be established from multiple detection results (e.g., multiple images). Finally, the system of equations may be solved, for example, using a linear least squares method or a Levenberg-Marquart (LM) algorithm.

[0074] In an exemplary manner, a neural network may be used to estimate seat parameters.

[0075] In this regard, the detected one or more seat components and / or seat component attributes are input into a neural network, and one or more seat parameters are received as outputs of the neural network.

[0076] It should be noted that the neural network used in this way is a trained neural network, which is trained accordingly, i.e. with respect to the provided input (data) and the expected output (data). For training, training data can be collected and used, the training data defining seat components and / or seat component attributes as inputs and target seat parameters as outputs. For example, the training data can include multiple data sets showing data of corresponding one or more seat components and / or seat component attributes and data of target seat parameters, such as one or more of seat track position, seat recline angle and seat headrest position, for a specific seat position (i.e., the actual (3D) position (or spatial configuration) of the seat, including one or more of the seat headrest, seat back, seat cushion and seat belt buckle). For training, real data can be set and used, the real data defining seat parameters for a predetermined spatial configuration (i.e., the actual (3D) position) of the seat in the vehicle cabin. For example, training data with seats placed in various (predetermined) seat constellations can be collected, and the relevant seat parameters can be stored as real data. The neural network can be trained using the training data and / or real data, for example as described above.

[0077] Alternatively, the detected geometric features of one or more seat components and / or seat component properties are input into a neural network, and one or more seat parameters are received as output of the neural network.

[0078] It should be noted that the neural network as used herein as such is a trained neural network, which is trained accordingly, i.e. with respect to the provided inputs (data) and the desired outputs (data). The training data used to train the neural network defines the detected geometric features of one or more seat components and / or seat component attributes as inputs and the target seat parameters as outputs. The geometric features may be and can be determined / encoded as described above with respect to geometric modeling, and therefore reference is made to the details accordingly.

[0079] Notwithstanding the foregoing, it should be noted that there are various possible (algorithmic) implementations for converting detected seat component attributes or geometric features determined / encoded thereon into one or more target seat parameters.

[0080] In an exemplary manner, a lookup table may be used to estimate the seat parameters.

[0081] In this regard, the detected one or more seat components and / or seat component properties are used as parameters to derive one or more seat parameters from a lookup table.

[0082] The lookup table so used may be calibrated and / or preconfigured (e.g., in an online or offline manner) such that the lookup table defines an association between at least one seat parameter and one or more seat component attributes (e.g., specific points or locations of a seat component detected in one or more images).

[0083] Figure 2E is a schematic diagram showing a grid cube representing a lookup table (LUT) for estimating one or more seat parameters. In this example, the upper left portion of the backrest (point / position), the upper left portion of the headrest (point / position), and the seat belt buckle (point / position) are used as parameters. The grid represents discrete steps of the LUT, which represents at least one seat parameter to be estimated / derived, such as seat track position and / or seat recline angle.

[0084] For example, calibration and / or pre-configuration of a look-up table (LUT) may be achieved by the following steps or operations.

[0085] First, discretize the space in which the three free seat parameters (e.g., seat track position, seat recline angle, headrest position) can vary (between corresponding minimum and maximum values). For example, the seat track position can vary between -25 cm (as the forward position or minimum value) and +25 cm (as the rearward position or maximum value) in steps of 5 cm (relative to the normal at 0 cm), for a total of 10 steps, the seat recline angle can vary between -30° (as the full forward position or minimum value) and +60° (as the full rearward position or maximum value) in steps of 10° (relative to the normal at 0°), for a total of 10 steps, and the headrest position can vary between -10 cm (as the lowest position or minimum value) and +10 cm (as the highest position or maximum value) in steps of 2 cm (relative to the normal at 0 cm), for a total of 10 steps.

[0086] Second, various constellations of seat parameters may be recorded and may be annotated or marked where N specific seat components are visible in one or more images. In this regard, specific points or locations of detected seat components may be considered. For example, the upper left portion of the seat headrest (point / location), the upper left portion of the seat back (point / location), and the (point / location) of the seat belt buckle may serve as parameters. For example, N may be 3, such that 3 instances of each of the upper left portion of the seat headrest (point / location), the upper left portion of the seat back (point / location), and the (point / location) of the seat belt buckle may serve as parameters.

[0087] Third, a discrete step size LUT, i.e., LUT value, is calculated. The LUT value holds / represents one or more seat parameters to be estimated. For example, the LUT value may hold / represent where the seat headrest and seat belt buckle will be projected into one or more images, i.e., at which pixel position.

[0088] Figure 3 is a flow chart showing a method according to at least one embodiment, which method may be considered or referred to as a method of facilitating vehicle operation or a vehicle operation facilitation method and / or a method of operating a vehicle device or an operation method (of a vehicle device).

[0089] like Figure 3 As shown, the method includes estimating one or more seat parameters of a seat in a vehicle cabin on which a person sits (block 31), and detecting one or more body parts of the person from one or more images of at least a portion of the vehicle cabin, and detecting body part attributes for each body part (block 32). Figure 1 The method shown is based on one or more images of at least a portion of a vehicle cabin and may therefore include inputting or receiving an image (although this is not Figure 3 For details of these images or captured images, refer to Figure 1 Description of the method shown.

[0090] It should be noted that you can use Figure 1 The method shown in FIG. 1 is used to estimate one or more seat parameters. Thus, block 31 may include Figure 1 If so, the same one or more images may be used in the operation / function of block 31 and the operation / function of block 32. In addition, the operation / function of block 32 need not be performed only after the operation / function of blocks 11 to 13 is completed, but may be performed at least partially simultaneously with the operation / function of blocks 12 and / or 13, for example.

[0091] like Figure 3 As shown, the method further includes controlling vehicle equipment based on the estimated one or more seat parameters and the detected one or more body part attributes of the person (block 33 ).

[0092] In the method, the vehicle equipment can be controlled in any feasible way. Figure 3 As shown, vehicle equipment may be controlled by adjusting a setting or configuration of the vehicle equipment (block 33 - 1 ) and / or causing a predetermined operation of the vehicle equipment (block 33 - 2 ).

[0093] It should be noted that the vehicle equipment may be controlled in response to the determination / identification (satisfaction) of one or more conditions, criteria, etc. (e.g., exceeding or failing to reach a predetermined threshold). Any such condition, criteria, etc. may depend on one or more estimated seat parameters and / or one or more detected body part attributes and / or one or more other parameters. As an illustrative but non-limiting example, in response to identifying that the seat track position (i.e., the seat position) is less than a given distance (threshold) from an airbag position (e.g., the surface of a glove box), the control may include outputting a warning to the passenger and / or automatically adjusting the seat track position so that the seat position is above the given distance (threshold) of the airbag position (which protects the passenger in the event of an airbag being triggered). As another illustrative but non-limiting example, in response to identifying that the vertical position of the seat headrest exceeds a given distance (threshold) to the head of the passenger (i.e., the person sitting on the corresponding seat), the control may include outputting a warning to the passenger and / or automatically adjusting the seat headrest so that the vertical position of the seat headrest is less than a given distance (threshold) to the head of the passenger (which protects the passenger in the event of a rear-end collision). Many more use cases are conceivable, only some of which are outlined below as illustrative and non-limiting examples.

[0094] The vehicle equipment may include, for example, one or more of a seat, a seat component of a seat, a controller, an airbag, and an acoustic and / or visual output device. The body part of a person may include, for example, one or more of a head and a torso of the person. One or more body part attributes may include, for example, one or more of the following: a bounding box of a detected body part, wherein the bounding box of the body part surrounds the body part in one or more images; one or more key points of the detected body part, wherein the one or more key points of the body part define feature points of the body part in one or more images; and an orientation of the detected body part, wherein the orientation of the body part defines a spatial orientation of the body part in the vehicle cabin.

[0095] When the vehicle device is at least one seat component of the seat, controlling the vehicle device may include, for example, adjusting at least one seat component so that seat parameters corresponding to the at least one seat component are adapted to one or more detected body part attributes of a person sitting on the seat.

[0096] It should be noted that although Figure 3 is a flowchart showing a method, but it can also be regarded as a block diagram showing a corresponding device or its internal structure / components. That is, it is configured to perform Figure 3 The apparatus of the method shown may include means, modules, components, circuits, etc. for performing any one or more of the operations / functions shown in blocks 31 , 32 , and 33 .

[0097] In the following, examples and / or details of the above-mentioned operations / functions of detecting body parts and related body part attributes are described.

[0098] like Figures 4 to 6 As shown, in one or more images (e.g., in image frames), specific parts of a (human) body can be detected. For any of the body parts detected in this way, at least one body part attribute (group) is detected (as a basis for subsequent vehicle device control). In other words, any body part attribute can be detected optically or in other words in an image-based manner and is relevant or effective for supporting the control of vehicle devices.

[0099] Detection of body parts and body part attributes can be achieved in several ways, such as any type of object detection. For example, such optical or image-based (object) detection can be performed or assisted by various systems, concepts or methods, such as convolutional neural networks, such as Region-based Convolutional Neural Network (R-CNN), Fast R-CNN, YOLO (You Only Look Once), etc.

[0100] Figure 4An exemplary detection result is shown, in which a bounding box (as an example of a body part attribute) around the face (as an example of a body part) of each of two persons is detected relative to the persons sitting in the two front seats. In this figure, the bounding boxes generated by the seat attribute detection (e.g., operation / function 12) (which are the bounding boxes of the seat headrests of the two front seats, i.e., around them) are represented by 41A and 41B, and the bounding boxes generated by the body attribute detection (e.g., operation / function 32) (which are the bounding boxes of the faces of the two persons, i.e., around them) are represented by 42A and 42B.

[0101] Figure 5 An exemplary detection result is shown, in which key points (as examples of body part attributes) on the face and torso (as examples of body parts) of each of the two persons are detected relative to the persons sitting in the two front seats. In this figure, the bounding box generated by the seat attribute detection (e.g., operation / function 12) (which is the bounding box of the seat headrests of the two front seats, i.e., around it) is represented by 51A and 51B, while the key points generated by the body attribute detection (e.g., operation / function 32) (which are the facial key points of the two persons, i.e., on it) are represented by 52A and 52B, and the key points generated by the body attribute detection (e.g., operation / function 32) (which are the torso key points of the two persons, i.e., on it) are represented by 53A and 53B. For further processing, a key point set, a (2D or 3D) center (e.g., center of gravity) of a key point, any feature representation of a key point set, etc. can be used.

[0102] Figure 6 An exemplary detection result is shown, in which the orientation (as an example of a body part attribute) of each of two persons' heads (as an example of a body part) is detected relative to the persons sitting in the two front seats. In this figure, the orientations (i.e., the orientations of the heads of two persons) resulting from the body property detection (such as operation / function 32) are represented by 62A and 62B. As shown herein, the orientations may be given / depicted in the axis directions of a 3D (image) coordinate system. It should be noted that for clarity, Figure 6 Not shown in FIG. 1 are bounding boxes resulting from seat attribute detection (such as operation / function 12). Although not shown, such bounding boxes may also exist in the corresponding images, such as bounding boxes for (i.e., surrounding) the seat headrests of the two front seats, as shown in FIG. Figure 4 41A and 41B in Figure 5 As shown in 51A and 51B.

[0103] It should be noted that one or more body part attributes may be detected in each image or image frame, such as any one or more of the exemplarily illustrated facial bounding box (see 42A, 42B), facial key points (see 52A, 52B), torso key points (see 53A, 53B), and head orientation (see 62A, 62B).

[0104] Although Figures 4 to 6 Detection results are shown for two people sitting in the front seats, but similar detection results can be obtained for any seats in the image, including the front seat on the left hand side of the image and one or more of any of the rear seats and the people sitting in the seats, as long as the seats and people (related body parts) are shown in the image. Detection results for any number of seats and people can be obtained. Although in Figures 4 to 6 Any of the detection results shown in shows only similar / same types of body part attributes (i.e., bounding box only, key points only, orientation only), but detection results with mixed types of body part attributes (e.g., face bounding box and face key points, face bounding box and head orientation) can be obtained.

[0105] Hereinafter, examples and / or details of the above-mentioned operations / functions of controlling vehicle equipment are described.

[0106] As mentioned above, body part attributes, e.g. Figures 4 to 6 The bounding boxes, key points and / or orientations shown are used as the basis for vehicle device control. This can be achieved in several ways.

[0107] For illustrative purposes, refer to Figures 4 to 6 , explains the use of a seat headrest for (automatically) adjusting the person sitting in the respective seat, i.e. the person's size, posture etc. In case of a rear-end collision, an optimally adjusted seat headrest (i.e. adjusted to the occupant) may avoid serious injuries. In order to determine whether the seat headrest is well adjusted to the occupant and to automatically adjust the seat headrest if desired / needed, additional detection of the person occupying the seat, i.e. the above-mentioned body part attributes, may be used. Depending on various criteria (e.g. desired accuracy), one or more types of body part attributes (e.g. a bounding box around the face (see Figure 4 )), which can be 2D / 3D keypoints of the face and / or torso (see Figure 5 ) and head orientation (see Figure 6 ).

[0108] Using a bounding box around the face of the person in the corresponding seat, the optimal seat headrest position can be determined relative to the facial position. Figure 4As shown, when / by comparing the various bounding boxes (eg, their relative (2D) vertical positions, their overlap, etc.), the facial position can be grasped relative to the seat headrest position.

[0109] Similarly, facial key points can be used to determine whether the seat headrest is well adjusted and, if desired / needed, determine the optimal seat headrest position relative to the facial position. Facial key points can be used to determine the seat headrest position relative to the person's head. Figure 5 As shown, when / by comparing the corresponding bounding box of the seat headrest position and the facial key points (e.g. their relative (2D) vertical positions, their overlap, etc.), the facial position can be grasped relative to the seat headrest position. Additional torso key points can be used to determine the body posture of the person, which can be used in addition to determine the optimal seat headrest position. If the person is sitting in a "normal" (upright) position, the optimal seat headrest position can be calculated relative to the person's head. If the person is not sitting in a "normal" (upright) position, but for example in a bent or lying position, it may be impossible or disadvantageous to calculate the optimal seat headrest position relative to the person's head, and a corresponding warning or indication can be presented to the person (e.g. visually and / or acoustically).

[0110] Identifying or having identified the head orientation of a person (e.g., in addition to the person's face / head position) can improve the calculation of the optimal seat headrest position. Identifying or having identified the 3D position of the seat headrest and the person and / or his or her face and / or torso key points can further enhance the accuracy of seat headrest adjustment. Figure 6 It is apparent that the optimal seat head restraint position (at least in terms of inclination and / or position in the vehicle / longitudinal direction) depends on the posture of the head. Therefore, information about the person's head orientation (e.g. the degree of forward tilt) can be appropriately applied to determine the optimal seat head restraint position relative to the person's head position.

[0111] Therefore, the optimal position of the seat headrest can be adjusted (i.e., its setting or configuration). In addition, the optimal position of the seat headrest so adjusted can be established, for example, by operating at least one (electric) motor to move the seat headrest to the optimal position so adjusted (i.e., a predetermined operation can be caused).

[0112] Nonetheless, it should be noted that there are various possible use cases and / or implementations for controlling vehicle devices based on seat parameters of a seat and body part attributes of a person sitting in the seat.

[0113] Typically, as described above, certain seat parameters are associated with different safety, comfort, and / or personalization related tasks. Thus, the techniques described herein may be applied in the context of any such tasks. Thus, the target vehicle device and its target control (e.g., adjusting its settings or configurations and / or causing its predetermined operation) may differ depending on the task in question.

[0114] For example, the correct positioning of seats and their (free) seat parameters (e.g., seat track position, seat recline angle, seat headrest position) in a vehicle contributes to several aspects of safe driving. For example, (1) identifying or having identified how close the front seats are to the airbag can be used to deploy the airbag differently or warn the occupant (i.e., the person sitting in the respective seat), e.g., when the seats are too close, (2) identifying or having identified the vertical position of the seat headrest relative to the occupant's (i.e., the person sitting in the respective seat) head can be used to warn the occupant or automatically adjust the seat headrest to optimally protect the occupant in the event of a rear-end collision, (3) identifying or having identified the seat track position and seat recline angle of a seat can support the estimation of a person's height using a 2D camera, which in turn can be effective, e.g., to distinguish between adults and children to appropriately deploy the airbag and adjust compliance, etc.

[0115] For example, the positioning of seats in a vehicle, including their (free) seat parameters such as seat track position, seat recline angle, seat headrest position, is relevant in the context of autonomous driving. For example, in the context of e.g. a Level 3 takeover situation (i.e. a situation where the driver will take over control in e.g. a Level 3 autonomous driving mode), identifying or having identified the seat recline angle of the driver's seat can be used to estimate the takeover time, i.e. the time it takes for the driver to get into a position to enable active driving. In this case, the seat recline angle can be adjusted automatically when it is expected that the driver will take over control again soon, or the driver can be warned if the estimated takeover time is too long (e.g. exceeds a predetermined threshold).

[0116] For example, the positioning of the seats in the vehicle, including their (free) seat parameters (e.g. seat track position, seat recline angle, seat headrest position) can be used for comfort and / or personalization purposes. When the seat track position and the seat recline angle (and seat headrest position) are or have been identified, rearview mirrors, such as exterior rearview mirrors, can be appropriately adjusted in view of the thereby expected position of the driver's head.

[0117] In vehicles with electronically adjustable seats, the estimated seat parameters may be used to automatically position the seat accordingly, e.g., for optimal visibility, passenger safety, passenger comfort, passenger personalization, etc. In vehicles with manually adjustable seats, the estimated seat parameters may be used to provide warnings or real-time feedback signals to passengers, e.g., to notify passengers, prompt passengers to perform corresponding adjustments, etc.

[0118] It should be noted that the methods described herein can be performed when the vehicle is stationary or turned off, but can also be performed when the vehicle is turned on or even during driving. The methods described herein can be triggered, started or performed regularly, such as when a predetermined time elapses, or on an event basis, such as when a certain event occurs.

[0119] Figure 7 is a schematic diagram showing an exemplary interior of a vehicle with possible camera locations.

[0120] The vehicle may include only one camera or multiple cameras located at different locations. With two or more cameras, a depth image or a 3D image may be created. The camera may be color or black and white, and these cameras, infrared cameras, depth cameras, thermal cameras, or a combination thereof may be placed, for example, in the middle of the front windshield or even in the middle of the rearview mirror, as shown in position 71. Additionally or alternatively, the camera or another camera may be located below the rearview mirror, as shown in position 72. If one camera is located at position 73, the other will typically also be located at position 74, but this is not mandatory. Additionally or alternatively, the camera or another camera may be located in the dashboard or center console, as shown in position 75. Each of positions 91 to 95 may also include two cameras co-located to achieve a 3D view of the interior of the vehicle.

[0121] The camera can capture images, for example, at regular time intervals or if triggered by an application that needs to estimate seat parameters and / or control vehicle equipment, as described herein. The application using the images can be executed on the vehicle computing system, or at least partially remotely, such as in the cloud. The result of the application can not only output a warning or control signal, but also trigger a display on the vehicle's main display 76 in the center console. The vehicle's main display can also be located in another location, such as at the dashboard behind the steering wheel.

[0122] Figure 8 800 is a schematic diagram showing an exemplary representation of a computing system. The schematic diagram can be understood to represent the internal components of a computing system 800 that implements one or more methods or operations / functions as described herein. The computing system 800 can be located in a vehicle and includes at least one processor 801 (e.g., a central processing unit (CPU)), a user interface 802, a network interface 803, and a main memory 806 that communicate with each other via a bus 805. Optionally, the computing system 800 may also include a static memory 807 and a disk drive unit (not shown), which also communicate with each other via the bus 805. A video display, an alphanumeric input device, and a cursor control device can be provided as examples of the user interface 802.

[0123] In addition, the computing system 800 may also include a camera interface 804 for communicating with a vehicle's onboard camera. Alternatively, the computing system 800 may communicate with the camera via a network interface 803. As described above, the camera is used to capture images. The computing system 800 may also be connected to a database system (not shown) via the network interface 803, wherein the database system stores at least a portion of the images required to provide the methods or operations / functions described herein.

[0124] The main memory 806 can be a random access memory (RAM) and / or any other volatile memory. The main memory 806 can store program codes for the sensing module 808 and / or the operating module 809, which implement the method or operation / function as described herein. Other modules required for other operations or functions described herein can be stored in the memory 806. The memory 806 can also store additional program data 810 required to provide the method or operation / function as described herein or other methods or operations / functions. Parts of the program data 810, the sensing module 808 and / or the operating module 809 can also be stored in a separate, for example, cloud storage, and at least partially executed remotely. For example, according to the method described herein, the main memory 806 can store one or more data about the detected seat components and seat component attributes, estimated seat parameters, detected body parts and body part attributes, etc. in the cache 811.

[0125] According to one embodiment, a vehicle is provided. The methods described herein may be stored as modules or program codes 808, 809, or 810 and may be at least partially included by the vehicle. Portions of the modules or program codes 808, 809, or 810 may also be stored on a cloud server and executed on the cloud server to reduce the computational workload on the computing system 800 of the vehicle. The vehicle may also include one or more cameras, for example, connected via a camera interface 804, for capturing one or more images.

[0126] According to one embodiment, any of the methods described herein, such as Figure 1 and Figure 3 The methods shown can be implemented as computerized methods or computer-implemented methods. For example, these methods can be implemented in software such as an application and / or can be executed by software such as an application.

[0127] According to one embodiment, a computer program including instructions is provided. When the computer program is executed by a computer, these instructions cause the computer to perform any of the methods described herein. The program code embodied in any system described herein can be distributed individually or collectively as a program product in a variety of different forms. In particular, the program code can be distributed using a computer-readable storage medium having computer-readable program instructions thereon, the computer-readable program instructions being used to cause a processor to perform aspects or features of the technology described herein.

[0128] Computer-readable storage media, which are non-transitory in nature, may include volatile and nonvolatile, as well as removable and non-removable tangible media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data. Computer-readable storage media may also include random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state memory technology, portable compact disk read-only memory (CD-ROM) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be read by a computer.

[0129] The computer-readable storage medium itself should not be interpreted as a transient signal (e.g., a radio wave or other propagating electromagnetic wave, an electromagnetic wave propagating through a transmission medium such as a waveguide, or an electrical signal transmitted through a wire). Computer-readable program instructions can be downloaded from the computer-readable storage medium to a computer, another type of programmable data processing device, or another device, or downloaded to an external computer or external storage device via a network.

[0130] It should be understood that although specific embodiments and variations are described herein, further modifications and substitutions are obvious to those skilled in the relevant art. Specifically, examples are provided by illustrating the principles of the present disclosure, and multiple specific examples, methods and devices are provided for making these principles effective.

[0131] In some embodiments, the operations / functions and / or actions specified in the flow chart, sequence diagram and / or block diagram may be reordered, processed serially and / or (at least partially) concurrently without departing from the scope of the present disclosure. In addition, any flow chart, sequence diagram and / or block diagram may include more or fewer blocks than those shown in some embodiments of the present disclosure.

[0132] The terms used herein are used only for the purpose of describing specific embodiments and are not intended to limit the present disclosure. It should also be understood that when used in this specification, the terms "include" and / or "comprise" specify the presence of the features, elements, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, features, elements, steps, operations, elements, components and / or combinations thereof. In addition, to the extent that the terms "include", "have", "contain", "consist of" or variations thereof are used in the detailed description or claims, these terms are intended to be included in a manner similar to the term "comprise".

[0133] Although the description of various embodiments has illustrated the technology of the present disclosure, and although these embodiments have been described in considerable detail, it is not intended to restrict or in any way limit the scope of the appended claims to such details. Additional advantages and modifications will be apparent to those skilled in the art. Therefore, the present disclosure in its broader aspects is not limited to the specific details, representative devices and methods, and / or illustrative examples shown and described herein. Therefore, the described embodiments should be understood as being provided by way of example for the purpose of teaching the general features and principles of the present disclosure, and should not be understood as limiting its scope, which is defined by the appended claims.

[0134] The present disclosure provides techniques for image-based vehicle cabin sensing and vehicle device operation. Such means include a method comprising: detecting one or more seat components of at least one seat in the vehicle cabin from one or more images of at least a portion of the vehicle cabin, and detecting seat component attributes of each detected seat component; and estimating one or more seat parameters of at least one seat based on the detected one or more seat component attributes, the one or more seat parameters of the seat indicating a spatial configuration of the seat in the vehicle cabin.

Claims

1. A method for facilitating vehicle operation, the method comprising the steps of: detecting one or more seat components of at least one seat in the vehicle cabin from one or more images of at least a portion of the vehicle cabin, and detecting seat component properties of each detected seat component; as well as Based on the detected one or more seat component properties, one or more seat parameters of the at least one seat are estimated, the one or more seat parameters of a seat being indicative of a spatial configuration of the seat in the vehicle cabin.

2. The method according to claim 1, wherein: The seat component properties include: a detected bounding box of a seat component, said bounding box of a seat component surrounding said seat component in said one or more images, and / or A specific point or location of the seat component is detected in the one or more images.

3. The method according to claim 1 or 2, wherein: The one or more seat parameters are estimated using geometric modeling or a neural network or a lookup table.

4. The method according to claim 3, wherein: Using geometric modeling to estimate the one or more seat parameters comprises the following steps: determining a geometric feature of a relationship between at least a portion of one or more detected seat component attributes and / or one or more groups of at least two detected seat component attributes; and Based on the determined geometric characteristics, the one or more seat parameters are calculated.

5. The method according to claim 4, wherein: The geometric features include one or more of a position of a seat component attribute, a size of a seat component attribute, a shape of a seat component attribute, a relative two-dimensional, horizontal and / or vertical position between two or more seat component attributes, and / or Calculating the one or more seat parameters utilizes a linear least squares method or a Levenberg-Marquart algorithm.

6. The method according to claim 3, wherein: Using a neural network to estimate the one or more seat parameters comprises the following steps: inputting the detected one or more seat component attributes into the neural network; and The one or more seat parameters are received as an output of the neural network.

7. The method according to claim 3, wherein: Using a lookup table to estimate the one or more seat parameters comprises the following steps: The one or more seat parameters are obtained from the lookup table using the detected one or more seat component properties as parameters.

8. The method according to any one of claims 1 to 7, wherein: The seat components of the seat include one or more of a seat cushion, a seat back, a seat headrest and a seat belt buckle, and / or The seat parameters of the seat include one or more of a seat track position, a seat recline angle, and a seat headrest position.

9. The method according to any one of claims 1 to 8, wherein: estimating one or more seat parameters of a seat in a cabin of the vehicle in which a person is seated, the method further comprising the steps of: detecting one or more body parts of the person from the one or more images, and detecting body part attributes of each body part; and The vehicle equipment is controlled based on the estimated one or more seat parameters and the detected one or more body part attributes of the person.

10. The method according to claim 9, wherein: Controlling the vehicle equipment includes one or more of the following steps: adjusting settings or configurations of the vehicle equipment; and A predetermined operation of the vehicle equipment is caused.

11. The method according to claim 9 or 10, wherein: The vehicle equipment includes one or more of the seat, a seat component of the seat, a controller, an airbag, and an acoustic and / or visual output device, and / or The body part of the person includes one or more of a head and a torso of the person.

12. The method according to any one of claims 9 to 11, wherein: The body part attributes include one or more of the following: a bounding box of a detected body part, the bounding box of the body part surrounding the body part in the one or more images, one or more key points of the detected body part, the one or more key points of the body part defining feature points of the body part in the one or more images, and A detected orientation of a body part, the orientation of a body part defining a spatial orientation of the body part in the vehicle cabin.

13. An apparatus configured to perform the method according to any one of claims 1 to 12.

14. A system, comprising: an image capture device configured to capture one or more images of at least a portion of a vehicle cabin, the vehicle cabin including at least one seat in the vehicle cabin; as well as The device according to claim 13.

15. A computer program product comprising instructions which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 12.

Images