Object class determination method, apparatus, device, and storage medium

Abstract

The application provides an object category determination method and device, equipment and storage medium, relates to the technical field of image processing, and is used for accurately identifying the main driving position and the vice driving position and guaranteeing the driving safety of the vehicle. The method comprises the following steps: capturing a snapshot image comprising a first type of reference component of a target vehicle and at least one passenger object by a snapshot device, determining the steering position attribute information of the target vehicle, then determining the offset angle of the target vehicle corresponding to the offset degree of the shooting angle of the snapshot device relative to the reference angle through the pose feature of the target vehicle in the snapshot image, mapping the first type of reference component and the at least one passenger object to the feature map corresponding to the reference angle through the offset angle, determining the relative position relationship between each passenger object and the first type of reference component according to the feature map, and finally determining the category of each passenger object through the steering position attribute information of the target vehicle and the obtained relative position relationship.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to an object category determination method and device, equipment and a storage medium. BACKGROUND

[0002] During driving, the driving safety is evaluated by monitoring the behavior of the driver on the main driving position to identify whether the driver has dangerous driving behaviors such as using a mobile phone, not wearing a seat belt or sleeping, so as to ensure the safety of vehicle driving. The behavior requirements for the passenger on the co-driver position are relatively low, so it is necessary to accurately distinguish the main driving position and the co-driver position from the captured image, and the accurate distinction of the main driving position and the co-driver position plays a very key role in safe driving evaluation.

[0003] At present, the main driving position and the co-driver position are distinguished by the left / right steering attribute of the vehicle or according to the position of the steering wheel, that is, in the captured image of the left steering vehicle, the seat on the right side of the window is the main driving position, and the seat on the left side is the co-driver position. If the image contains a steering wheel, the seat where the steering wheel is located is the main driving position. However, in the case of vehicle circulation between different regions, and the existence of left and right steering vehicles, or due to the installation conditions or installation scene of the monitoring device, the monitoring device is installed at a low height or has a large shooting angle, which may cause visual deviation of the left and right seats in the captured image, or the steering wheel is invisible, and so on. Such as the above-mentioned various situations will result in the inability to accurately identify the main driving position and the co-driver position. Therefore, the accuracy of the current way of distinguishing the main driving position and the co-driver position is not high. SUMMARY

[0004] The embodiments of the present application provide an object category determination method, device, equipment and storage medium to solve the technical problem of low accuracy of distinguishing the main driving position and the co-driver position in the prior art.

[0005] In one aspect, an object category determination method is provided, the method comprising:

[0006] Based on a snapshot image of a target vehicle captured by a snapshot device, determining the steering position attribute information of the target vehicle, the snapshot image comprising a first type of reference component of the target vehicle and at least one driving and riding object;

[0007] Based on the pose feature of the target vehicle in the snapshot image, determining the offset angle corresponding to the target vehicle, the offset angle representing the offset degree of the shooting angle of the snapshot device relative to the reference angle;

[0008] Based on the offset angle, mapping the first type of reference component and the at least one driving and riding object to a feature map corresponding to the reference angle;

[0009] determine, respectively, a first relative position relationship between each of the at least one passenger object and the first reference component based on the feature map;

[0010] determine a category of each of the at least one passenger object based on the rudder position attribute information and the obtained first relative position relationship.

[0011] In one aspect, an object category determination apparatus is provided, and the apparatus comprises:

[0012] a first determination unit configured to determine rudder position attribute information of a target vehicle based on a snapshot image of the target vehicle captured by a snapshot device, the snapshot image comprising a first reference component of the target vehicle and at least one passenger object;

[0013] a second determination unit configured to determine an offset angle corresponding to the target vehicle based on a pose feature of the target vehicle in the snapshot image, the offset angle representing a degree of offset of a capturing perspective of the snapshot device relative to a reference perspective;

[0014] a mapping unit configured to map the first reference component and the at least one passenger object into a feature map corresponding to the reference perspective based on the offset angle;

[0015] a positioning unit configured to determine, respectively, a first relative position relationship between each of the at least one passenger object and the first reference component based on the feature map;

[0016] a classification unit configured to determine a category of each of the at least one passenger object based on the rudder position attribute information and the obtained first relative position relationship.

[0017] Optionally, the first determination unit is specifically configured to:

[0018] determine whether the snapshot image comprises a windshield wiper of the target vehicle;

[0019] if yes, extract a pose feature of the windshield wiper from the snapshot image, and determine the rudder position attribute information based on the pose feature;

[0020] if no, determine the rudder position attribute information based on regional information of an area where the snapshot device is located.

[0021] Optionally, the second determination unit is specifically configured to:

[0022] determine at least two offset reference points based on feature points of at least two second reference components of the target vehicle in the snapshot image;

[0023] determine the offset angle corresponding to the target vehicle based on a relative positional relationship between the at least two offset reference points in the snapshot image and a relative positional relationship between the at least two offset reference points in the reference view angle.

[0024] Optionally, the second determining unit is specifically configured to:

[0025] construct a reference coordinate system, wherein a line connecting the at least two offset reference points is parallel to a longitudinal axis of the reference coordinate system in the reference view angle;

[0026] determine the offset angle based on an included angle between the line connecting the at least two offset reference points and the longitudinal axis in the shooting view angle.

[0027] Optionally, the mapping unit is specifically configured to:

[0028] determine a target mapping relationship from a plurality of mapping relationships based on a preset included angle range in which the offset angle is located, wherein each preset included angle range uniquely corresponds to one mapping relationship in the plurality of mapping relationships;

[0029] determine second position information of a component feature point of the first type of reference component in the feature map based on first position information of the component feature point in the snapshot image and the target mapping relationship;

[0030] determine fourth position information of the at least one driver or passenger object in the feature map based on third position information of the at least one driver or passenger object in the snapshot image and the target mapping relationship.

[0031] Optionally, the mapping unit is specifically configured to:

[0032] when the offset angle is greater than the preset included angle, determine a first mapping relationship as the target mapping relationship; or

[0033] when the offset angle is less than the preset included angle, determine a second mapping relationship as the target mapping relationship.

[0034] wherein the shooting view angle of the snapshot device in the first mapping relationship is different from that in the second mapping relationship.

[0035] Optionally, the positioning unit is specifically configured to:

[0036] determine a first sub-region and a second sub-region from a region of the feature map in which the first type of reference component is located;

[0037] for each of the driver or passenger objects, the following operations are performed:

[0038] For one driving object, if a first coincidence degree of the driving object and the first sub-region is greater than a second coincidence degree of the driving object and the second sub-region, it is determined that the driving object is located in the first sub-region;

[0039] If the first coincidence degree of the driving object and the first sub-region is less than the second coincidence degree of the driving object and the second sub-region, it is determined that the driving object is located in the second sub-region.

[0040] Optionally, the classification unit is specifically used for:

[0041] Based on the rudder position attribute information, the first sub-region and the second sub-region are determined to correspond to respective driving region categories;

[0042] For each driving object, the following operations are respectively performed:

[0043] For one driving object, if the driving object is located in the first sub-region, the category of the driving object is determined according to the driving region category of the first sub-region;

[0044] If the driving object is located in the second sub-region, the category of the driving object is determined according to the driving region category of the second sub-region.

[0045] Optionally, the first determination unit is specifically used for:

[0046] It is determined whether the snapshot image only includes one driving object;

[0047] If not, it is determined whether the snapshot image includes a steering wheel of the target vehicle;

[0048] If not, based on a snapshot image of the target vehicle captured by a snapshot device, rudder position attribute information of the target vehicle is determined.

[0049] Optionally, the first determination unit is specifically used for:

[0050] It is determined whether a ratio between a region range occupied by a first type of reference component of the target vehicle and a region range occupied by the target vehicle in the snapshot image is greater than a preset proportion threshold;

[0051] If yes, based on a snapshot image of the target vehicle captured by a snapshot device, rudder position attribute information of the target vehicle is determined.

[0052] In one aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above methods.

[0053] In an aspect, a computer storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement steps of any of the above methods.

[0054] In an aspect, a computer program product or a computer program is provided, which includes computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute steps of any of the above methods.

[0055] The beneficial effects of the embodiments of the present application are as follows:

[0056] In the embodiments of the present application, a snapshot image including a first type of reference component of a target vehicle and at least one occupant is captured by a snapshot device, the rudder position attribute information of the target vehicle is determined, and then the offset angle of the target vehicle corresponding to the offset degree of the shooting angle of the snapshot device relative to the reference angle is determined through the pose feature of the target vehicle in the snapshot image. The first type of reference component and the at least one occupant are mapped into the feature map corresponding to the reference angle through the offset angle, the relative position relationship between each occupant and the first type of reference component is determined according to the feature map, and finally the category of each occupant is determined through the rudder position attribute information of the target vehicle and the obtained relative position relationship. The method corrects the deviation of the shooting angle of the snapshot device relative to the reference angle through the offset angle, eliminates the visual deviation of the positions of the main driver seat and the assistant driver seat in the snapshot image, and then combines the rudder position attribute of the target vehicle, so as to accurately identify the category of the main driver seat or the assistant driver seat to which each occupant belongs, so as to evaluate the corresponding safe driving behavior of different occupants and ensure the driving safety of the vehicle.

[0057] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.

[0059] Figure 1 An application scenario schematic diagram is provided for the embodiments of the present application.

[0060] Figure 2A flowchart of an object category determination method provided by an embodiment of the present application is shown in FIG. 1.

[0061] Figure 3a A partial schematic diagram of a snapshot image under a reference view angle provided by an embodiment of the present application is shown in FIG. 2.

[0062] Figure 3b A partial schematic diagram of a snapshot image under a shooting view angle provided by an embodiment of the present application is shown in FIG. 3.

[0063] Figure 3c A partial schematic diagram of a feature map after offset correction provided by an embodiment of the present application is shown in FIG. 4.

[0064] Figure 4 A schematic diagram of a reference coordinate system under a reference view angle provided by an embodiment of the present application is shown in FIG. 5.

[0065] Figure 5 A schematic diagram of an offset angle under a shooting view angle provided by an embodiment of the present application is shown in FIG. 6.

[0066] Figure 6 A schematic diagram of a relative position relationship between a driver or passenger and a first reference component provided by an embodiment of the present application is shown in FIG. 7.

[0067] Figure 7 A flowchart of another object category determination method provided by an embodiment of the present application is shown in FIG. 8.

[0068] Figure 8 A structural schematic diagram of an object category determination apparatus provided by an embodiment of the present application is shown in FIG. 9.

[0069] Figure 9 A structural schematic diagram of a computer device provided by an embodiment of the present application is shown in FIG. 10. DETAILED DESCRIPTION

[0070] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application. In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other at will. Moreover, although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

[0071] The design idea of the embodiments of the present application will be briefly introduced below.

[0072] In a traffic monitoring scenario, a monitoring camera can identify the behavior of the driver and passenger in the captured vehicle image by analyzing the captured vehicle image, perform relevant driving behavior safety evaluation, and determine whether there is dangerous behavior that violates the safety driving regulations. Especially for the driver on the main driving position, it is necessary to accurately monitor whether there is dangerous driving behavior such as playing mobile phone, not wearing safety belt or sleeping, in order to ensure the safety of the traffic road. The behavior requirement for the passenger on the co-driver position is relatively low, and it is usually only necessary to identify whether there is behavior that interferes with the driving of the vehicle. Therefore, accurately distinguishing the different drivers and passengers on the main driving position and the co-driver position from the captured image plays a very key role in subsequent safety driving evaluation.

[0073] At present, the main driving position and the co-driver position are mainly distinguished by the left / right steering attribute of the vehicle or according to the position of the steering wheel. That is, in the captured image of the left steering vehicle, the seat on the right side of the window is the main driving position, and the seat on the left side is the co-driver position. If there is a steering wheel in the image, the seat where the steering wheel is located is the main driving position. However, when facing the situation that vehicles circulate among various regions and there are left and right steering vehicles at the same time, it will be difficult to determine the steering attribute of the photographed vehicle. On the other hand, due to the limitation of the installation condition or installation scene of the monitoring device, when the monitoring device is installed at a low height or at a large shooting angle, the steering wheel in the captured vehicle image may not be visible, so it is not possible to determine the main driving position according to the position of the steering wheel. Or the captured vehicle image may have a large vehicle angle, and the vehicle may occupy a large part of the image, which will cause visual deviation of the left and right seats, and is very unfavorable for subsequent image recognition. Such as the above-mentioned situations, will all lead to the inability to accurately identify the main driving position and the co-driver position. Therefore, the accuracy of the current method for identifying the main driving position and the co-driver position is not high.

[0074] In view of the above problems, the embodiment of the present application provides a method for determining the category of an object. The method comprises the following steps: capturing a snapshot image including a first type of reference component of a target vehicle and at least one passenger object by a snapshot device; determining the steering position attribute information of the target vehicle; determining the offset angle of the target vehicle corresponding to the offset degree of the shooting angle of the snapshot device relative to the reference angle through the pose feature of the target vehicle in the snapshot image; mapping the first type of reference component and the at least one passenger object to the feature map corresponding to the reference angle through the offset angle; determining the relative position relationship between each passenger object and the first type of reference component according to the feature map; and finally determining the category of each passenger object through the steering position attribute information of the target vehicle and the obtained relative position relationship. The method corrects the deviation of the shooting angle of the snapshot device relative to the reference angle through the offset angle, eliminates the visual deviation of the position of the driver seat and the front passenger seat in the snapshot image, and then combines the steering position attribute of the target vehicle, so as to accurately identify the category of the driver seat or the front passenger seat to which each passenger object belongs, so as to evaluate the corresponding safe driving behavior of different passenger objects and ensure the driving safety of the vehicle.

[0075] In order to further improve the accuracy and efficiency of the identification of the driver and the front passenger, the embodiment of the present application further determines whether there is only one passenger object in the snapshot image, so as to quickly identify the driver seat. Otherwise, according to whether the snapshot image includes a steering wheel, the main driver area where the steering wheel is located and the corresponding main driver are quickly determined. When the above two conditions are not met, the pose feature of the wiper of the target vehicle is used to determine the steering position attribute, so as to more accurately determine the steering position attribute information of the target vehicle when there are left steering and right steering vehicles at the same time, and facilitate the subsequent identification process of the driver and the front passenger.

[0076] In order to solve the technical problem that when the shooting angle of the snapshot device deviates from the reference angle, the positions of the main and vice driver seats in the snapshot image are visually deviated, and the recognition accuracy of the main and vice drivers is affected, the embodiments of the present application consider that the positions of the vehicle components on the vehicle body structure are relatively fixed, and the offset reference point is determined by the feature points of the multiple reference components of the target vehicle, so that the abstract offset degree can be accurately quantified by the position change of the offset reference point in the shooting angle and the reference angle, so as to subsequently correct the image by the offset angle. Specifically, the embodiments of the present application further convert the different positions of the offset reference point in the shooting angle and the reference angle into an included angle which can be quantified by an angle, and more accurately calculate the offset degree of the shooting angle relative to the reference angle by the offset angle, thereby further improving the accuracy of the main and vice driver recognition. After the size of the offset angle is determined, the embodiments of the present application determine the different mapping relationships by determining the different offset directions of the shooting angle relative to the reference angle through the size of the offset angle, so as to accurately map the first type of reference component and the driver and passenger objects to the feature map corresponding to the reference angle, correct the angle of view offset, and accurately determine the relative position relationship between each driver and passenger object and the first type of reference component in the reference angle through the corrected position information in the feature map, so as to accurately recognize the main and vice driver seats.

[0077] In order to better understand the above technical solutions, the technical solutions of the present application will be described in detail below through the drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, and are not limitations of the technical solutions of the present application. In the case of no conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0078] The technical solutions provided by the embodiments of the present application can be applied to various vehicle main and vice driver seat recognition scenarios, such as Figure 1 As shown in the figure, an application scenario provided by the embodiments of the present application can include a vehicle 100, a snapshot device 110, an object category determination device 120 and a network 130.

[0079] The vehicle 100 can be any vehicle that can be driven on the road, and the driver compartment contains a main driver seat and a vice driver seat. The relative position relationship between the main and vice driver seats in the driver compartment is determined according to the steering position attribute of the vehicle, that is, the main driver seat of the left steering vehicle is located on the left side of the driver compartment, and the main driver seat of the right steering vehicle is located on the right side of the driver compartment, and the vice driver seat is opposite to the main driver seat.

[0080] The snapshot device 110 is a shooting device capable of collecting image data of a target such as a vehicle and a driver, and can transmit the captured image data to the object category determination device for subsequent image recognition, including but not limited to one or more of a fast ball camera, a gun camera, a gimbal camera, a half-sphere camera, a long-focus, medium-focus or wide-angle camera, etc.

[0081] The object category determination device 120 is a computing device with certain computing power and capable of realizing image recognition function, which is the execution subject of the object category determination method provided in the embodiments of the present application, i.e., the object category determination device can obtain the vehicle snapshot image captured by the snapshot device, and realize the function of identifying the driver and the front passenger based on the object category determination method provided in the embodiments of the present application. It should be understood that the computing device provided in the embodiments of the present application can be a terminal device or a server or other device with computing function. In a possible implementation, the object category determination device can be the same device as the snapshot device 110, i.e., the snapshot device can perform the identification of the driver and the front passenger based on the captured vehicle image data by itself, or it can be a server connected to the snapshot device, the snapshot device transmits the image data to the server through a network, the server receives the image data and realizes the function of identifying the driver and the front passenger based on the object category determination method provided in the embodiments of the present application.

[0082] The object category determination device 120 and the snapshot device 110 can be connected through a network 130, which can be a wired network or a wireless network. For example, the wireless network can be a mobile cellular network such as a fourth generation (4G) network, a fifth generation (5G) network or a new radio (NR) network, or a wireless fidelity (WIFI) network, and of course can also be other possible networks, which are not limited in the embodiments of the present application.

[0083] In a possible implementation, when the computing power of the snapshot device allows, the object category determination device 120 can also be realized by the snapshot device 110, i.e., the object category determination device 120 can be the snapshot device 110.

[0084] It should be noted that, Figure 1 The number of vehicles and the number of snapshot devices shown are only illustrative, and in fact, the number of vehicles and the number of snapshot devices are not limited, which are not specifically limited in the embodiments of the present application. And Figure 1 The components and structures shown are only exemplary and not limiting, and in actual scenarios, other components and structures can also be provided as needed.

[0085] Of course, the method provided by the embodiments of the present application is not limited to the application scenarios shown Figure 1 The application scenarios shown can also be used in other possible application scenarios, and the embodiments of the present application do not limit the same. The functions that can be achieved by the various devices in the application scenarios shown will be described together in subsequent method embodiments, and will not be described in detail here. Figure 1 The functions that can be achieved by the various devices in the application scenarios shown will be described together in subsequent method embodiments, and will not be described in detail here.

[0086] The object category determination method provided by the exemplary embodiments of the present application will be described below in combination with the application scenarios described above and with reference to the accompanying drawings. It should be noted that the application scenarios described above are only shown to facilitate understanding of the spirit and principles of the present application, and the embodiments of the present application are not limited in this respect.

[0087] Referring to Figure 2 FIG. 1 shows a flowchart of the object category determination method provided by the embodiments of the present application, which is exemplarily described taking the object category determination device as the execution subject. The specific implementation process of the method is as follows:

[0088] Step 201: Determine the rudder position attribute information of the target vehicle based on the snapshot image of the target vehicle captured by the snapshot device.

[0089] In the embodiments of the present application, after the snapshot device captures the snapshot image containing the target vehicle, the snapshot image is sent to the object category determination device, so that the object category determination device can analyze the snapshot image to determine the rudder position attribute information of the target vehicle and perform subsequent identification of the main and auxiliary drivers. The rudder position attribute information of the vehicle represents the rudder position type of the vehicle. The main driver position of vehicles of different rudder position types is located at different positions in the vehicle. Generally, the main driver position of a left-hand drive vehicle is located on the left side of the driver's cabin, and the main driver position of a right-hand drive vehicle is located on the right side of the driver's cabin.

[0090] In one possible implementation, the object category determination device can determine the rudder position attribute information by the pose features of the wiper of the target vehicle in the snapshot image. If the wiper does not exist in the snapshot image, the rudder position attribute information is determined by the regional information of the area where the snapshot device is located.

[0091] Specifically, because the main driver positions of left-hand drive or right-hand drive vehicles are different, the wiper arm will inevitably block the view of the higher position when it swings in front of the driver. In order to facilitate the driver to observe the situation outside the vehicle, the installation position and pointing direction of the wiper of the left-hand drive or right-hand drive vehicle are different, for example, even if the wiper is not started and is in a flat state, the wiper of the left-hand drive vehicle is installed on the left side of the window, and the wiper arm points to the right side, while the wiper of the right-hand drive vehicle is installed on the right side of the window, and the wiper arm points to the left side. Therefore, the rudder position attribute of the target vehicle can be determined according to the different pose features of the wiper of the target vehicle.

[0092] Specifically, if the wiper of the target vehicle in the snapshot image is invisible due to the snapshot angle and other factors, and the rudder position attribute of the vehicle cannot be accurately determined according to the wiper, the rudder position attribute of the target vehicle in the same region can be determined according to the regional information of the region where the snapshot device is located, and the rudder position attribute of most or even all vehicles in the region. In addition, the rudder position attribute can also be determined by any other possible way, and the embodiments of the present application do not limit this.

[0093] In one possible way, the object category determination device can also determine whether the snapshot image is image data that can be used to determine the object category according to whether the ratio between the area range occupied by the first type of reference component of the target vehicle in the snapshot image and the area range occupied by the target vehicle is greater than a preset proportion threshold before determining the rudder position attribute of the target vehicle.

[0094] Specifically, when the target vehicle is not facing the camera of the snapshot device, the side or even the tail of the target vehicle may occupy most of the area of the snapshot image, which makes such snapshot image unable to reflect the specific situation of the driver and the front passenger in the vehicle, and is not suitable for driver and front passenger identification. Therefore, the embodiments of the present application first determine the image as image data that can be used to determine the object category through the first type of reference component of the target vehicle, such as the size of the image area occupied by the front window and the size of the image area occupied by the entire vehicle. When the ratio is greater than a preset proportion threshold, the subsequent driver and front passenger identification process is performed.

[0095] Step 202: determining the offset angle corresponding to the target vehicle based on the pose feature of the target vehicle in the snapshot image.

[0096] In the embodiments of the present application, the object category determination device determines the offset degree of the shooting angle of the snapshot device relative to the reference angle, that is, the offset angle, by analyzing the pose feature of the target vehicle in the snapshot image.

[0097] Specifically, taking the front window as the first type of reference component of the target vehicle as an example, the shooting angle of the snapshot device facing the front of the target vehicle is the reference angle, and the reference Figure 3a As shown in the reference angle, the driver and the front passenger in the front window in the snapshot image will be symmetrically distributed on both sides of the center line of the front window, and the offset angle corresponding to the target vehicle is zero. As shown in FIG. 6B, the shooting angle of the snapshot device is not facing the front of the target vehicle, and the driver and the front passenger in the front window in the snapshot image are not symmetrically distributed on both sides of the center line of the front window, and the offset angle corresponding to the target vehicle is not zero. Figure 3bAs shown, once the shooting angle of the snapshot device is offset relative to the reference angle, the driver and the front passenger will no longer be symmetrically distributed on both sides of the center line of the front window. At this time, if the driver and the front passenger are determined by the center line of the front window, the recognition result of the driver and the front passenger will be inaccurate. Therefore, in order to accurately recognize the driver and the front passenger, the offset angle of the shooting angle of the snapshot device relative to the reference angle needs to be accurately determined, so as to correct the snapshot image by the offset angle, such as Figure 3c As shown, the corrected snapshot image can accurately distinguish the driver and the front passenger areas by the center line of the front window.

[0098] In a possible implementation, the object category determination device can determine a plurality of offset reference points by the feature points of a plurality of second type reference components of the target vehicle in the snapshot image. Then, according to the relative positional relationship of the plurality of offset reference points in the snapshot image and the relative positional relationship of the plurality of offset reference points under the reference angle, the offset angle corresponding to the target vehicle is determined.

[0099] Specifically, the second type reference component can be at least two of the commonly used fixed components such as the left and right side mirrors, the middle rearview mirror, the car logo, or the car head of the vehicle. The feature points of the component can use the center point of the image area where the component is located, or use the point position information such as the top left corner and the bottom right corner of the image area where the component is located determined by the target detection algorithm to determine the position information of the center point. Reference Figure 4 As shown, the positions of the above-mentioned second type reference components of most vehicles on the vehicle body structure are relatively fixed, and since the vehicle structure has symmetry, under the reference angle, the straight line where the feature points of the left and right side mirrors of the vehicle are located is perpendicular to the straight line where the feature points of the middle rearview mirror and the car logo (license plate) are located in the projection plane parallel to the plane where the snapshot image is located. Similarly, when the shooting angle of the snapshot device is offset relative to the reference angle, the offset angle relative to the reference angle can be determined according to the relative positional relationship between the left and right side mirrors and the middle rearview mirror and the car logo (license plate) of the vehicle.

[0100] In one possible implementation, the object category determination device can perform target detection on various reference parts of the target vehicle in the captured image using object detection algorithms, etc., to determine whether there are relevant reference parts in the captured image and their location information in the image region. Specifically, a pre-trained pre-defined convolutional neural network can be used to extract characteristic information of different vehicle parts in the image, such as Faster Region-based Convolutional Neural Networks (Faster R-CNN) or other algorithm models and techniques, such as object detection algorithms (You Only Look Once, YOLO), Single Shot Multi Box Detector (SSD), etc., or an algorithm model and technique composed of one or more of these techniques can be used.

[0101] In one possible implementation, the offset angle corresponding to the target vehicle can be determined by constructing a reference coordinate system and combining the different relative positional relationships of the offset reference point in the captured image and under the reference viewpoint.

[0102] For details, please refer to Figure 5 As shown, the feature points of the center rearview mirror and the car logo (or license plate) are used as offset reference points. The line connecting these offset reference points, or their parallel lines, is used as the vertical axis of the reference coordinate system from the reference viewpoint. Therefore, from the shooting viewpoint, the angle between the line connecting these offset reference points and the vertical axis is the offset angle corresponding to the target vehicle. The magnitude of the offset angle is then calculated using trigonometric functions. By constructing a coordinate system, the different positional relationships of the offset reference points under the shooting and reference viewpoints are transformed into angles quantified by angles, allowing for a more accurate calculation of the degree of offset between the shooting viewpoint and the reference viewpoint.

[0103] For details, please refer to the above. Figure 5 As shown, taking the center rearview mirror and the car logo as examples of the second type of reference components, based on the coordinates (X...) of point A at the top left corner of the image area where the center rearview mirror is located... A Y A ), the coordinates of point B in the lower right corner (X) B Y B Based on the principle of symmetry, the coordinates of point C, the center point of the image area containing the center rearview mirror, are: Similarly, based on the coordinates (X) of point D at the top left corner of the image area where the car logo is located... D Y D ), the coordinates of point E in the lower right corner (X E Y E From this, we can determine that the coordinates of point F, the center point of the image area containing the car logo, are... Therefore, according to the arctangent function, the angle θ between the line connecting the center points of the rearview mirror and the car logo and the vertical axis of the reference coordinate system is as follows:

[0104]

[0105] Step 203: Based on the offset angle, map the first type of reference component and at least one driving object to the feature map corresponding to the reference viewpoint.

[0106] In this embodiment of the application, after determining the offset angle, the object category determination device performs offset correction on the shooting angle of the captured image by mapping the first type of reference component and the driving object to the feature map corresponding to the reference viewpoint.

[0107] Specifically, taking the front window as the first type of reference component as an example, when the object category determination device performs offset correction, it uses the target mapping relationship to determine the second position information of the component feature points in the feature map based on the first position information of the component feature points in the captured image; and, based on the third position information of the driver and passenger object in the captured image, it uses the target mapping relationship to determine the fourth position information in the feature map.

[0108] Specifically, when determining the target mapping relationship corresponding to the offset angle, the rotation matrix corresponding to the offset angle along the vertical axis can be determined first according to the three-dimensional space transformation rules, as shown below:

[0109]

[0110] Because the target object (i.e., the image area where the first type of reference component and the driver / passenger object are located) has three-dimensional coordinates in the captured image. Its original three-dimensional coordinates from the reference perspective The transformation is derived from the rotation matrix along the vertical axis described above, as shown below:

[0111]

[0112] Since the recognition of the driver and passenger seats only requires considering their left-right positional relationship on the horizontal axis of the reference coordinate system, only the transformation of the X-coordinate needs to be considered. The transformation process is as follows:

[0113]

[0114] Since the image is in two-dimensional space, the influence of the depth Z-axis does not need to be considered, and the transformation process can be ignored. The influence of the value. Therefore, the position information of the target object mapped to the feature map corresponding to the reference viewpoint can be determined by the inverse transformation of the rotation matrix, as shown below:

[0115]

[0116] In one possible implementation, the object category determination device can determine the preset angle range where the offset angle is located based on the relative size relationship between the offset angle and the preset angle, wherein each preset angle range uniquely corresponds to one of a plurality of mapping relationships, and the target mapping relationship to be used is determined from the plurality of preset mapping relationships based on the preset angle range.

[0117] In one possible implementation, the different offset directions of the shooting angle of the capture device relative to the reference angle can be determined based on the different relative magnitudes of the offset angle and zero, thereby determining the corresponding target mapping relationship.

[0118] Specifically, taking the front window as the first type of reference component as an example, the first mapping relationship and the second mapping relationship corresponding to the deflection angle θ in different preset angle ranges are as follows:

[0119] (1) When θ>0, the first mapping relationship is:

[0120]

[0121] Among them, X G To capture the coordinates (X) of point G, the top left corner of the image area containing the front windshield in the image. G Y G The x-axis coordinate in ().

[0122] When the offset angle is greater than zero, the shooting angle of the capture device is tilted to the right relative to the reference angle. At this time, an inverse transformation is performed with the upper left corner of the first type of reference component as the origin. Based on the X-coordinate of the feature points of the first type of reference component or the vehicle / passenger object in the captured image, its position in the feature map is calculated through this mapping relationship. Coordinates are used to map the first type of reference component and the driving / riding object onto the feature map corresponding to the reference viewpoint.

[0123] (2) When θ < 0, the second mapping relationship is:

[0124]

[0125] Among them, as above Figure 5 As shown, X H The coordinates (X) of point H, the lower right corner of the image area containing the windshield. H Y H The x-axis coordinate of ).

[0126] When the offset angle is less than zero, the shooting angle of the capture device deflects to the left relative to the reference angle. At this time, the rotation is performed with the lower right corner of the first type of reference component as the origin. Based on the X-coordinate of the feature points of the first type of reference component or the vehicle / passenger object in the captured image, its position in the feature map is calculated through this mapping relationship. The coordinates map the first type of reference component and the driving object to the feature map corresponding to the reference viewpoint, so that the original captured image with offset error can be corrected.

[0127] Step 204: Based on the feature map, determine the first relative positional relationship between each driving object and the first type of reference component.

[0128] In this embodiment of the application, after mapping the first type of reference component and the driving / riding object to the feature map, the object category determination device can determine the first relative positional relationship between each driving / riding object and the first type of reference component based on the feature map.

[0129] In one possible implementation, the object category determination device may first determine a first sub-region and a second sub-region from the region of the feature map of the first type of reference component. Then, for each driving or riding object, the degree of overlap between it and the first and second sub-regions is determined, and the sub-region in which the driving or riding object is located is determined based on the relative magnitude of the degree of overlap.

[0130] Specifically, such as Figure 6 As shown, the object category determination device can determine the center line of the image region where the first type of reference component is located in the feature map based on the position information of each feature point of the first type of reference component in the feature map. Based on the center line, the image region where the first type of reference component is located is divided into two symmetrical and equal sub-regions, namely the first sub-region and the second sub-region. Then, for each driving / riding object in the feature map, if the first degree of overlap between the driving / riding object and the first sub-region is greater than its second degree of overlap with the second sub-region, then the driving / riding object is determined to be located in the first sub-region; otherwise, it is determined to be located in the second sub-region.

[0131] Step 205: Based on the steering position attribute information and the obtained first relative position relationships, determine the category of each driving and riding object.

[0132] In the embodiments of this application, the object category determination device can determine the driver and passenger categories of each driver and passenger after determining the steering position attribute information of the target vehicle and the first relative positional relationship between each driver and passenger and the first type of reference component.

[0133] In a possible implementation, the object category determining device can determine the first sub-region and the second sub-region of the first type of reference component according to the steering attribute information of the target vehicle, each corresponding to a respective driver / passenger region category. For example, when the first sub-region and the second sub-region are respectively located on the left and right sides of the image region where the first type of reference component is located, if the target vehicle is a left-hand drive vehicle, the first sub-region corresponds to the driver seat region of the target vehicle, and the second sub-region corresponds to the front passenger seat region of the target vehicle. If the target vehicle is a right-hand drive vehicle, the first sub-region corresponds to the front passenger seat region of the target vehicle, and the second sub-region corresponds to the driver seat region of the target vehicle. The object category determining device can determine the driver / passenger category of each driver / passenger object according to the driver / passenger region category of the sub-region where the driver / passenger object is located. For example, when the target vehicle is a right-hand drive vehicle, and the first sub-region and the second sub-region are respectively located on the left and right sides of the image region where the first type of reference component is located, if one driver / passenger object of the target vehicle is located in the first sub-region, it can be determined that the category of the driver / passenger object is the front passenger, and the category of the driver / passenger object located in the second sub-region is the driver.

[0134] In a possible implementation, in order to improve the driver / passenger recognition efficiency, the object category determining device can further determine, before determining the steering attribute information of the target vehicle, whether there is only one driver / passenger object in the snapshot image. If yes, it can be directly determined that the category of the driver / passenger object is the driver, and the driver seat is quickly recognized. If not, it is further determined whether the snapshot image includes the steering wheel of the target vehicle. If yes, the driver / passenger category of the driver / passenger object located in the driver seat region where the steering wheel is located is determined to be the driver. When neither of the above two conditions is met, the steering attribute information of the target vehicle is determined, and the subsequent driver / passenger recognition process is performed, to ensure the accuracy of driver / passenger recognition.

[0135] In a possible implementation, the reference Figure 7As shown, the object class determination device, after receiving the snapshot image of the target vehicle captured by the snapshot device, performs target detection on the vehicle components and the driver and passenger in the vehicle by using a target detection algorithm, and identifies the pose feature of the vehicle. After judging that the ratio of the front window of the target vehicle to the image area range occupied by the entire vehicle body is greater than a preset proportion threshold, it is determined that the snapshot image can be used for object class determination. According to the target detection algorithm result, it is judged whether there is only one driver and passenger object in the image. If not, it is judged whether the image includes the steering wheel of the target vehicle. If not, it is judged whether the image includes the wiper. If yes, the rudder position attribute information of the target vehicle is determined according to the pose feature of the detected wiper. Otherwise, it is determined according to the regional information of the region. According to the feature point coordinates of the vehicle middle mirror and the vehicle logo determined by the target detection, the offset angle corresponding to the target vehicle is calculated. According to the size of the offset angle, the position coordinates of the front window and the driver and passenger are corrected to obtain the relative position relationship of the front window and each driver and passenger in the feature map. And according to the determined rudder position attribute, the class of each driver and passenger is determined.

[0136] Please refer to Figure 8 Based on the same inventive concept, the embodiments of the present application also provide an object class determination device 80, which comprises:

[0137] The first determination unit 801 is configured to determine the rudder position attribute information of the target vehicle based on the snapshot image of the target vehicle captured by the snapshot device, the snapshot image comprising a first type of reference component of the target vehicle and at least one driver and passenger object.

[0138] The second determination unit 802 is configured to determine the offset angle corresponding to the target vehicle based on the pose feature of the target vehicle in the snapshot image, the offset angle representing the offset degree of the shooting angle of the snapshot device relative to the reference angle.

[0139] The mapping unit 803 is configured to map the first type of reference component and the at least one driver and passenger object to the feature map corresponding to the reference angle based on the offset angle.

[0140] The positioning unit 804 is configured to determine the first relative position relationship between each driver and passenger object and the first type of reference component respectively based on the feature map.

[0141] The classification unit 805 is configured to determine the class of each driver and passenger object based on the rudder position attribute information and the obtained first relative position relationship.

[0142] Optionally, the first determination unit 801 is specifically configured to:

[0143] determine whether the snapshot image includes the wiper of the target vehicle;

[0144] If yes, pose features of the windscreen wiper are extracted from the snapshot image, and the rudder position attribute information is determined based on the pose features.

[0145] If no, the rudder position attribute information is determined based on regional information of an area where the snapshot device is located.

[0146] Optionally, the second determining unit 802 is specifically configured to:

[0147] determine at least two offset reference points based on feature points of at least two second-type reference components of the target vehicle in the snapshot image;

[0148] determine the offset angle corresponding to the target vehicle based on a relative positional relationship of the at least two offset reference points in the snapshot image and a relative positional relationship of the at least two offset reference points under the reference visual angle.

[0149] Optionally, the second determining unit 802 is specifically configured to:

[0150] construct a reference coordinate system, wherein, under the reference visual angle, a line connecting the at least two offset reference points is parallel to a longitudinal axis of the reference coordinate system;

[0151] determine the offset angle according to an included angle between the line connecting the at least two offset reference points and the longitudinal axis under the shooting visual angle.

[0152] Optionally, the mapping unit 803 is specifically configured to:

[0153] determine the target mapping relationship from a plurality of mapping relationships based on a preset included angle range in which the offset angle is located, wherein each preset included angle range uniquely corresponds to one mapping relationship in the plurality of mapping relationships;

[0154] determine second positional information of the component feature point in the feature map based on the first positional information of the component feature point of the first-type reference component in the snapshot image and the target mapping relationship;

[0155] determine fourth positional information of the at least one driver or passenger in the feature map based on the third positional information of the at least one driver or passenger in the snapshot image and the target mapping relationship.

[0156] Optionally, the mapping unit 803 is specifically configured to:

[0157] when the offset angle is greater than a preset included angle, determine the first mapping relationship as the target mapping relationship; or,

[0158] when the offset angle is less than the preset included angle, determine the second mapping relationship as the target mapping relationship.

[0159] wherein the shooting visual angle of the snapshot device is different in the offset direction in the first mapping relationship and the second mapping relationship.

[0160] Optionally, the positioning unit 804 is specifically used for:

[0161] determining a first sub-region and a second sub-region in the region of the feature map from the first type of reference component;

[0162] for each driving object, the following operations are respectively performed:

[0163] for a driving object, if a first coincidence degree of the driving object and the first sub-region is greater than a second coincidence degree of the driving object and the second sub-region, it is determined that the driving object is located in the first sub-region;

[0164] if the first coincidence degree of the driving object and the first sub-region is less than the second coincidence degree of the driving object and the second sub-region, it is determined that the driving object is located in the second sub-region.

[0165] Optionally, the classification unit 805 is specifically used for:

[0166] determining a driving region category corresponding to each of the first sub-region and the second sub-region based on the steering attribute information;

[0167] for each driving object, the following operations are respectively performed:

[0168] for a driving object, if the driving object is located in the first sub-region, a category of the driving object is determined according to the driving region category of the first sub-region;

[0169] if the driving object is located in the second sub-region, a category of the driving object is determined according to the driving region category of the second sub-region.

[0170] Optionally, the first determination unit 801 is specifically used for:

[0171] determining whether the snapshot image includes only one driving object;

[0172] if not, determining whether the snapshot image includes a steering wheel of the target vehicle;

[0173] if not, determining the steering attribute information of the target vehicle based on the snapshot image of the target vehicle captured by the snapshot device.

[0174] Optionally, the first determination unit 801 is specifically used for:

[0175] determining whether a ratio between a region range occupied by the first type of reference component of the target vehicle and a region range occupied by the target vehicle in the snapshot image is greater than a preset proportion threshold;

[0176] if yes, determining the steering attribute information of the target vehicle based on the snapshot image of the target vehicle captured by the snapshot device.

[0177] By the above device, the first type of reference component including the target vehicle and at least one occupant in the snapshot image captured by the snapshot device is determined, the rudder position attribute information of the target vehicle is determined, and the offset angle of the target vehicle corresponding to the offset degree of the shooting angle of the snapshot device relative to the reference angle is determined through the pose feature of the target vehicle in the snapshot image. The first type of reference component and at least one occupant are mapped to the feature map corresponding to the reference angle through the offset angle, the relative position relationship between each occupant and the first type of reference component is determined according to the feature map, and finally the category of each occupant is determined through the rudder position attribute information of the target vehicle and the obtained relative position relationship. The method corrects the deviation of the shooting angle of the snapshot device relative to the reference angle through the offset angle, eliminates the visual deviation of the positions of the driver seat and the front passenger seat in the snapshot image, and combines the rudder position attribute of the target vehicle itself, so as to accurately identify the category of the driver seat or the front passenger seat to which each occupant in the vehicle belongs, so as to evaluate the corresponding safe driving behavior of different occupants and ensure the driving safety of the vehicle.

[0178] For the convenience of description, each part is divided into a unit module (or module) according to the function and is described respectively. Of course, the functions of each unit (or module) can be realized in the same or multiple software or hardware when implementing the present application. The device can be used to execute the method shown in each embodiment of the present application, and therefore, the functions of each functional module of the device and the like can be referred to the description of the foregoing embodiments, and will not be described in detail.

[0179] Please refer to Figure 9 , based on the same technical concept, the embodiments of the present application also provide a computer device. In one embodiment, the computer device as shown can include a memory 901, a communication module 903 and one or more processors 902.

[0180] The memory 901 is used to store the computer program executed by the processor 902. The memory 901 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, and the data storage area can store various operation instruction sets and the like.

[0181] Memory 901 may be volatile memory, such as random-access memory (RAM); memory 901 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 901 may be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 901 may be a combination of the above-described memories.

[0182] Processor 902 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 902 is used to implement the above-mentioned object category determination method when calling computer programs stored in memory 901.

[0183] The communication module 903 is used to communicate with the object category determination device or other network devices.

[0184] This application embodiment does not limit the specific connection medium between the memory 901, communication module 903, and processor 902 described above. This application embodiment... Figure 9 The memory 901 and the processor 902 are connected via a bus 904, which is in... Figure 9 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. The 904 bus can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 9 It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0185] The memory 901 stores a computer storage medium containing computer-executable instructions for implementing the object category determination method of the embodiments of this application. The processor 902 is used to execute the object category determination methods of the above embodiments.

[0186] Based on the same inventive concept, embodiments of this application also provide a storage medium storing a computer program, which, when executed on a computer, causes a computer processor to perform the steps in the object category determination method according to various embodiments of this application described above.

[0187] In some possible implementations, various aspects of the object category determination method provided in this application can also be implemented in the form of a program product, which includes program code. When the program product is run on a computer device, the program code is used to cause the computer device to perform the steps in the object category determination method according to the various exemplary embodiments of this application described above. For example, the computer device can perform the steps of the various embodiments.

[0188] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0189] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a computing device. However, the program product of this application is not limited thereto. In this application, the readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with a command execution system, apparatus, or device.

[0190] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0191] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0192] Program code for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0193] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0194] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0195] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0196] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0197] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for determining object categories, characterized in that, The method includes: Based on the captured image of the target vehicle taken by the capture device, the steering position attribute information of the target vehicle is determined. The captured image includes a first type of reference component of the target vehicle and at least one driver or passenger. Based on the pose features of the target vehicle in the captured image, the offset angle corresponding to the target vehicle is determined, and the offset angle represents the degree of offset of the shooting angle of the capture device relative to the reference angle. Based on the offset angle, the first type of reference component and the at least one driving / riding object are mapped onto the feature map corresponding to the reference viewpoint; Based on the feature map, the first relative positional relationship between each of the driving and riding objects and the first type of reference component is determined respectively; Based on the steering position attribute information and the obtained first relative position relationships, the respective categories of each driving and riding object are determined; The step of determining the first relative positional relationship between each of the driving / riding objects and the first type of reference component based on the feature map includes: From the region of the feature map, the first sub-region and the second sub-region are determined; For each of the aforementioned drivers and passengers, the following operations shall be performed: For a given driving / riding object, if the first degree of overlap between the driving / riding object and the first sub-region is greater than the second degree of overlap between the driving / riding object and the second sub-region, then the driving / riding object is determined to be located in the first sub-region. If the first degree of overlap between the driving / riding object and the first sub-region is less than the second degree of overlap between the driving / riding object and the second sub-region, it is determined that the driving / riding object is located in the second sub-region.

2. The method as described in claim 1, characterized in that, Determining the steering position attribute information of the target vehicle includes: Determine whether the captured image includes the windshield wipers of the target vehicle; If so, the position and pose features of the windshield wiper are extracted from the captured image, and the rudder position attribute information is determined based on the position and pose features; If not, the rudder position attribute information is determined based on the geographical information of the area where the capture device is located.

3. The method as described in claim 1, characterized in that, Determining the offset angle corresponding to the target vehicle based on the pose features of the target vehicle in the captured image includes: Based on feature points of at least two second-type reference components of the target vehicle in the captured image, at least two offset reference points are determined; Based on the relative positional relationship of the at least two offset reference points in the captured image, and the relative positional relationship of the at least two offset reference points under the reference viewpoint, the offset angle corresponding to the target vehicle is determined.

4. The method as described in claim 3, characterized in that, Before determining the offset angle corresponding to the target vehicle based on the relative positional relationship of the at least two offset reference points in the captured image and the relative positional relationship of the at least two offset reference points under the reference viewpoint, the method further includes: Construct a reference coordinate system; wherein, from the reference viewpoint, the line connecting the at least two offset reference points is parallel to the vertical axis of the reference coordinate system; Based on the relative positional relationship of the at least two offset reference points in the captured image, and the relative positional relationship of the at least two offset reference points under the reference viewpoint, the offset angle corresponding to the target vehicle is determined, including: The offset angle is determined based on the angle between the line connecting the at least two offset reference points and the vertical axis under the shooting angle.

5. The method as described in claim 1, characterized in that, The step of mapping the first type of reference component and the at least one driving / riding object onto the feature map corresponding to the reference viewpoint based on the offset angle includes: Based on the preset angle range in which the offset angle is located, a target mapping relationship is determined from multiple mapping relationships; wherein, each preset angle range uniquely corresponds to one of the multiple mapping relationships. Based on the first position information of the component feature points of the first type of reference component in the captured image, the second position information of the component feature points in the feature map is determined using the target mapping relationship; Based on the third position information of the at least one vehicle in the captured image, the fourth position information of the at least one vehicle in the feature map is determined using the target mapping relationship.

6. The method as described in claim 5, characterized in that, The determination of the target mapping relationship from multiple mapping relationships based on the preset angle range of the offset angle includes: When the offset angle is greater than the preset included angle, the first mapping relationship is determined as the target mapping relationship; or, When the offset angle is less than the preset included angle, the second mapping relationship is determined as the target mapping relationship; In the first mapping relationship and the second mapping relationship, the shooting angle of the capture device is offset from the reference angle in different directions.

7. The method as described in claim 1, characterized in that, The step of determining the category of each of the driving and riding objects based on the steering position attribute information and the obtained first relative position relationships includes: Based on the steering position attribute information, determine the driving and riding area categories corresponding to the first sub-region and the second sub-region, respectively. For each of the aforementioned drivers and passengers, the following operations shall be performed: For a driver / passenger, if the driver / passenger is located in the first sub-region, the category of the driver / passenger is determined according to the driving / passenger region category of the first sub-region. If the vehicle is located in the second sub-region, the category of the vehicle is determined according to the vehicle area category of the second sub-region.

8. The method as described in claim 1, characterized in that, The determination of the steering position attribute information of the target vehicle based on the captured image of the target vehicle by the capture device includes: Determine whether the captured image contains only one driver or passenger. If not, then determine whether the captured image includes the steering wheel of the target vehicle; If not, the steering position attribute information of the target vehicle is determined based on the captured image of the target vehicle taken by the capture device.

9. The method as described in claim 1, characterized in that, The determination of the steering position attribute information of the target vehicle based on the captured image of the target vehicle by the capture device includes: Determine whether the ratio between the area occupied by the first type of reference component of the target vehicle and the area occupied by the target vehicle in the captured image is greater than a preset ratio threshold. If the value is greater than the target vehicle's value, the steering position attribute information of the target vehicle is determined based on the captured image of the target vehicle taken by the capture device.

10. An object category determination device, characterized in that, include: The first determining unit is used to determine the steering position attribute information of the target vehicle based on the captured image of the target vehicle taken by the capture device. The captured image includes a first type of reference component of the target vehicle and at least one driver or passenger. The second determining unit is used to determine the offset angle corresponding to the target vehicle based on the pose features of the target vehicle in the captured image, wherein the offset angle represents the degree of offset of the shooting angle of the capturing device relative to the reference angle. A mapping unit is used to map the first type of reference component and the at least one driving object onto a feature map corresponding to the reference viewpoint based on the offset angle; The positioning unit is used to determine, based on the feature map, the first relative positional relationship between each of the driving and riding objects and the first type of reference component; A classification unit is used to determine the category of each of the driving and riding objects based on the steering position attribute information and the obtained first relative position relationships; Specifically, when the positioning unit determines the first relative positional relationship between each of the driving / riding objects and the first type of reference component based on the feature map, it is used for: From the region of the feature map, the first sub-region and the second sub-region are determined; For each of the aforementioned drivers and passengers, the following operations shall be performed: For a given driving / riding object, if the first degree of overlap between the driving / riding object and the first sub-region is greater than the second degree of overlap between the driving / riding object and the second sub-region, then the driving / riding object is determined to be located in the first sub-region. If the first degree of overlap between the driving / riding object and the first sub-region is less than the second degree of overlap between the driving / riding object and the second sub-region, it is determined that the driving / riding object is located in the second sub-region.

11. A computer device, characterized in that, include: At least one processor, and Memory connected to the at least one processor; The memory stores instructions executable by the at least one processor, which executes the method as described in any one of claims 1-9 by executing the instructions stored in the memory.

12. A readable storage medium, characterized in that, Including memory, The memory is used to store instructions that, when executed by a processor, cause a device including the readable storage medium to perform the method as described in any one of claims 1 to 9.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 9.

Images