Vehicle information determination method and device, electronic equipment and storage medium

Abstract

The present application provides a kind of vehicle information determination method, device, electronic equipment and storage medium, the method comprises: receiving the vehicle image data reported by image acquisition device;Based on vehicle image data and target detection algorithm, the boundary box of target vehicle and the boundary box of each identification component of target vehicle are determined in pixel coordinate system, and target vehicle is any one vehicle in monitoring area;Based on each boundary box, the sub-image corresponding to target vehicle and the sub-image corresponding to each identification component are extracted from vehicle image data;Based on vehicle detection model and the sub-image corresponding to target vehicle, the body color of target vehicle is determined, and based on vehicle detection model and the sub-image corresponding to each identification component, the identification information of each identification component is determined. Through vehicle detection model, each sub-image is detected, the body color and the identification information of each identification component can be determined, and the utilization rate of vehicle image data reported by image acquisition device can be improved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, electronic device, and storage medium for determining vehicle information. Background Technology

[0002] Artificial intelligence has been widely used in the Internet of Vehicles (IoV), where object detection technology can be used to detect target objects, such as vehicles, in images.

[0003] Roadside equipment in the Internet of Vehicles (IoV) is generally equipped with image acquisition devices. These devices capture images to be detected and upload them to a cloud platform. The cloud platform extracts features from the images to obtain image features, which are then input into a pre-trained license plate recognition model. The license plate recognition model is used to identify the vehicle's license plate information based on the image features. The cloud platform then queries the vehicle database based on the license plate information to obtain other attribute information of the vehicle (such as vehicle color or vehicle brand). If the license plate information is not found in the vehicle database, other attribute information of the vehicle cannot be obtained.

[0004] In related technologies, cloud platforms can only identify license plate information from vehicle image data reported by image acquisition devices. How to improve the utilization rate of vehicle image data reported by image acquisition devices is an urgent problem to be solved in the industry. Summary of the Invention

[0005] To address the problems existing in the prior art, embodiments of the present invention provide a method, apparatus, electronic device, and storage medium for determining vehicle information.

[0006] In a first aspect, the present invention provides a method for determining vehicle information, comprising:

[0007] Receive vehicle image data reported by the image acquisition device;

[0008] Based on the vehicle image data and target detection algorithm, the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle are determined in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area.

[0009] Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data;

[0010] Based on the vehicle detection model and the sub-image corresponding to the target vehicle, the body color of the target vehicle is determined, and based on the vehicle detection model and the sub-image corresponding to each identification component, the identification information of each identification component is determined.

[0011] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle image data includes a first vehicle image, a second vehicle image, a third vehicle image, and a fourth vehicle image. The first vehicle image is an image captured by a first camera at a first moment, the second vehicle image is an image captured by a second camera at the first moment, the third vehicle image is an image captured by the first camera at a second moment, and the fourth vehicle image is an image captured by the second camera at the second moment. The second moment is a moment after the first moment, and there is an interval of one acquisition cycle between the first moment and the second moment. The image acquisition device includes a binocular camera, which includes a first camera and a second camera.

[0012] The step of determining the bounding box of the target vehicle and the bounding boxes of each identifying component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm includes:

[0013] Based on the target detection algorithm, the first bounding box and the second bounding box corresponding to the target vehicle in the first vehicle image and the second vehicle image, respectively, are determined; and based on the target detection algorithm, the bounding box of each first vehicle in the third vehicle image is determined; and based on the target detection algorithm, the bounding box of each second vehicle in the fourth vehicle image is determined.

[0014] Based on the DeepSORT algorithm and the first bounding box, the third bounding box of the target vehicle is determined in the bounding boxes of each first vehicle; and based on the DeepSORT algorithm and the second bounding box, the fourth bounding box of the target vehicle is determined in the bounding boxes of each second vehicle.

[0015] Based on the target detection algorithm, the first bounding box, the second bounding box, the third bounding box, and the fourth bounding box, the fifth bounding box of each identification component of the target vehicle in the first vehicle image, the sixth bounding box of each identification component of the target vehicle in the second vehicle image, the seventh bounding box of each identification component of the target vehicle in the third vehicle image, and the eighth bounding box of each identification component of the target vehicle in the fourth vehicle image are determined.

[0016] Optionally, according to a vehicle information determination method provided by the present invention, after determining the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, the method further includes:

[0017] The first pixel coordinate position of the target vehicle is determined based on the first bounding box; the second pixel coordinate position of the target vehicle is determined based on the second bounding box; the third pixel coordinate position of the target vehicle is determined based on the third bounding box; and the fourth pixel coordinate position of the target vehicle is determined based on the fourth bounding box.

[0018] Based on camera parameters, the first pixel coordinate position, and the second pixel coordinate position, the first world coordinate position of the target vehicle is determined; and based on the camera parameters, the third pixel coordinate position, and the fourth pixel coordinate position, the second world coordinate position of the target vehicle is determined, wherein the camera parameters are parameters determined by calibrating the binocular camera;

[0019] The speed of the target vehicle is determined based on the first world coordinate position, the second world coordinate position, and the acquisition period.

[0020] Optionally, according to a vehicle information determination method provided by the present invention, the step of extracting sub-images corresponding to the target vehicle and sub-images corresponding to each identifier component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each identifier component includes:

[0021] Based on the first bounding box, a first sub-image of the target vehicle is extracted from the first vehicle image; and based on the second bounding box, a second sub-image of the target vehicle is extracted from the second vehicle image; and based on the third bounding box, a third sub-image of the target vehicle is extracted from the third vehicle image; and based on the fourth bounding box, a fourth sub-image of the target vehicle is extracted from the fourth vehicle image.

[0022] After extracting the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component, the method further includes:

[0023] Based on the camera parameters, the first sub-image, and the second sub-image, a first depth map of the target vehicle is obtained; and based on the camera parameters, the third sub-image, and the fourth sub-image, a second depth map of the target vehicle is obtained.

[0024] Based on the first depth map and the camera position relationship, a first volume is determined; and based on the second depth map and the camera position relationship, a second volume is determined, wherein the camera position relationship is the positional relationship between the first camera and the second camera;

[0025] The volume of the target vehicle is determined based on the first volume and the second volume.

[0026] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a license plate detection sub-model, the identification component of the target vehicle includes the target license plate, and the step of determining the identification information of each identification component based on the vehicle detection model and the sub-images corresponding to each identification component includes:

[0027] Based on the license plate detection sub-model and the sub-image corresponding to the target license plate, the first recognition information of the target license plate is determined;

[0028] The license plate detection sub-model is obtained by training a CRNN model based on the license plate dataset.

[0029] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a vehicle logo detection sub-model, the identification component of the target vehicle includes a target vehicle logo, and the step of determining the identification information of each identification component based on the vehicle detection model and the sub-images corresponding to each identification component includes:

[0030] Based on the vehicle logo detection sub-model and the sub-image corresponding to the target vehicle logo, the recognition information of the target vehicle logo is determined;

[0031] The vehicle logo detection sub-model is obtained by training a CNN model based on the vehicle logo dataset.

[0032] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a vehicle body color detection sub-model, and the step of determining the vehicle body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle includes:

[0033] Based on the vehicle body color detection sub-model and the sub-image corresponding to the target vehicle, the vehicle body color of the target vehicle is determined;

[0034] The vehicle color detection sub-model is obtained by training a MobileNet model based on the vehicle color dataset.

[0035] Secondly, the present invention also provides a vehicle information determining device, comprising:

[0036] The receiving module is used to receive vehicle image data reported by the image acquisition device;

[0037] The first determining module is used to determine the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, wherein the target vehicle is any vehicle within the monitoring area.

[0038] The extraction module is used to extract sub-images corresponding to the target vehicle and sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component.

[0039] The second determining module is used to determine the body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle, and to determine the identification information of each identification component based on the vehicle detection model and the sub-image corresponding to each identification component.

[0040] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle information determination method as described above.

[0041] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle information determination method as described above.

[0042] The vehicle information determination method, apparatus, electronic device, and storage medium provided by this invention detect vehicle image data reported by an image acquisition device using a target detection algorithm. This allows for the determination of the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle. Based on these bounding boxes, sub-images corresponding to the target vehicle and each identification component can be extracted from the vehicle image data. Furthermore, a vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component. This enables the determination of the vehicle color information of the target vehicle and the identification information of each identification component, thereby improving the utilization rate of the vehicle image data reported by the image acquisition device. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating the vehicle information determination method provided by the present invention;

[0045] Figure 2 This is a schematic diagram of the vehicle information determination device provided by the present invention;

[0046] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0048] Figure 1 This is a flowchart illustrating the vehicle information determination method provided by the present invention, as shown below. Figure 1 As shown, the vehicle information determination method can be executed by an electronic device, such as a server. The method includes:

[0049] Step 101: Receive vehicle image data reported by the image acquisition device.

[0050] Specifically, any vehicle within the monitoring area of ​​the image acquisition device can be taken as the target vehicle. In order to determine multiple attribute information of the target vehicle, vehicle image data reported by the image acquisition device can be received, and then the vehicle image data can be analyzed and processed.

[0051] Step 102: Based on the vehicle image data and the target detection algorithm, determine the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area.

[0052] Specifically, after acquiring vehicle image data, the vehicle image data reported by the image acquisition device can be detected by a target detection algorithm, which can determine the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system.

[0053] It's understandable that an image is composed of pixels, and pixel coordinates are the position of a pixel within the image. Pixel coordinates can be determined using a pixel coordinate system. Generally, a pixel coordinate system, uv, is established with the top-left corner of the image as the origin and uses pixels as the unit. Here, the x-coordinate u and y-coordinate v of a pixel are its column number and row number in the image array, respectively.

[0054] Optionally, the identification components of the target vehicle may include the license plate and the vehicle logo. By using a target detection algorithm to detect the vehicle image data reported by the image acquisition device, the bounding box of the target license plate and the bounding box of the target vehicle logo can be determined in the pixel coordinate system.

[0055] Step 103: Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data.

[0056] Specifically, after determining the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle, the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component can be extracted from the vehicle image data based on the bounding boxes.

[0057] It is understandable that the vehicle images reported by the image acquisition device may include multiple vehicles within the monitored area. The vehicle images reported by the image acquisition device can be cropped based on the bounding box to obtain sub-images. The sub-image corresponding to the target vehicle may only include the target vehicle, and the sub-image corresponding to the identification component may only include a single identification component. The sub-image may only include a single object to be detected, so that the vehicle detection model can accurately identify the object to be detected in the sub-image during subsequent processing.

[0058] Optionally, the identification components of the target vehicle may include the target license plate and the target vehicle logo. By detecting the vehicle image data reported by the image acquisition device using a target detection algorithm, the bounding box of the target license plate and the bounding box of the target vehicle logo can be determined in the pixel coordinate system. Then, the sub-image corresponding to the target license plate and the bounding box of the target vehicle logo can be extracted from the vehicle image data based on the bounding boxes.

[0059] Step 104: Based on the vehicle detection model and the sub-image corresponding to the target vehicle, determine the body color of the target vehicle, and based on the vehicle detection model and the sub-image corresponding to each identification component, determine the identification information of each identification component.

[0060] Specifically, after extracting the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component, the vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component, thereby determining the vehicle body color of the target vehicle and the recognition information of each identification component.

[0061] Optionally, the vehicle detection model may include: a license plate detection sub-model, a vehicle logo detection sub-model, and a vehicle body color detection sub-model.

[0062] Optionally, the identification component of the target vehicle may include the target license plate, and the sub-image corresponding to the target license plate can be detected by a license plate detection sub-model to determine the license plate recognition information of the target vehicle.

[0063] Optionally, the identification component of the target vehicle may include the target vehicle logo, and the logo detection sub-model can be used to detect the sub-image corresponding to the target vehicle logo in order to determine the logo recognition information of the target vehicle.

[0064] Optionally, the vehicle body color can be determined by detecting the sub-image corresponding to the target vehicle through a vehicle body color detection sub-model.

[0065] The vehicle information determination method provided by this invention detects vehicle image data reported by an image acquisition device using a target detection algorithm. This determines the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle. Based on the bounding boxes, sub-images corresponding to the target vehicle and each identification component can be extracted from the vehicle image data. Furthermore, a vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component. This method can determine the vehicle body color of the target vehicle and the identification information of each identification component, thereby improving the utilization rate of vehicle image data reported by the image acquisition device.

[0066] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle image data includes a first vehicle image, a second vehicle image, a third vehicle image, and a fourth vehicle image. The first vehicle image is an image captured by a first camera at a first moment, the second vehicle image is an image captured by a second camera at the first moment, the third vehicle image is an image captured by the first camera at a second moment, and the fourth vehicle image is an image captured by the second camera at the second moment. The second moment is a moment after the first moment, and there is an interval of one acquisition cycle between the first moment and the second moment. The image acquisition device includes a binocular camera, which includes a first camera and a second camera.

[0067] The step of determining the bounding box of the target vehicle and the bounding boxes of each identifying component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm includes:

[0068] Based on the target detection algorithm, the first bounding box and the second bounding box corresponding to the target vehicle in the first vehicle image and the second vehicle image, respectively, are determined; and based on the target detection algorithm, the bounding box of each first vehicle in the third vehicle image is determined; and based on the target detection algorithm, the bounding box of each second vehicle in the fourth vehicle image is determined.

[0069] Based on the DeepSORT algorithm and the first bounding box, the third bounding box of the target vehicle is determined in the bounding boxes of each first vehicle; and based on the DeepSORT algorithm and the second bounding box, the fourth bounding box of the target vehicle is determined in the bounding boxes of each second vehicle.

[0070] Based on the target detection algorithm, the first bounding box, the second bounding box, the third bounding box, and the fourth bounding box, the fifth bounding box of each identification component of the target vehicle in the first vehicle image, the sixth bounding box of each identification component of the target vehicle in the second vehicle image, the seventh bounding box of each identification component of the target vehicle in the third vehicle image, and the eighth bounding box of each identification component of the target vehicle in the fourth vehicle image are determined.

[0071] Specifically, in order to extract sub-images of the target vehicle and sub-images of each identification component from multiple vehicle images, the image acquisition device can be equipped with a binocular camera, which includes a first camera and a second camera. The vehicle image data reported by the image acquisition device can include images captured by the first camera and images captured by the second camera. The vehicle image data can include a first vehicle image captured by the first camera at a first moment, a second vehicle image captured by the second camera at a first moment, a third vehicle image captured by the first camera at a second moment, and a fourth vehicle image captured by the second camera at a second moment.

[0072] Specifically, after acquiring the first vehicle image, the second vehicle image, the third vehicle image, and the fourth vehicle image, the first vehicle image can be detected using an object detection algorithm to obtain the first bounding box of the target vehicle; the second vehicle image can be detected using an object detection algorithm to obtain the second bounding box of the target vehicle; the third vehicle image can be detected using an object detection algorithm to determine the bounding box of each first vehicle in the third vehicle image; and the fourth vehicle image can be detected using an object detection algorithm to determine the bounding box of each second vehicle in the fourth vehicle image.

[0073] Specifically, the DeepSORT algorithm can be used to track the target vehicle. Based on the first bounding box, the bounding box corresponding to the target vehicle in the third vehicle image can be tracked. That is, the third bounding box of the target vehicle is determined in the bounding boxes of each first vehicle. Based on the second bounding box, the bounding box corresponding to the target vehicle in the fourth vehicle image can be tracked. That is, the fourth bounding box of the target vehicle is determined in the bounding boxes of each second vehicle.

[0074] Specifically, the image within the first bounding box can be identified based on the object detection algorithm to determine the fifth bounding box of each identification component of the target vehicle in the first vehicle image; the image within the second bounding box can be identified based on the object detection algorithm to determine the sixth bounding box of each identification component of the target vehicle in the second vehicle image; the image within the third bounding box can be identified based on the object detection algorithm to determine the seventh bounding box of each identification component of the target vehicle in the third vehicle image; and the image within the fourth bounding box can be identified based on the object detection algorithm to determine the eighth bounding box of each identification component of the target vehicle in the fourth vehicle image.

[0075] Understandably, after determining the bounding box of the target vehicle and the sub-images of each identifying component for multiple vehicle images, the sub-images of the target vehicle and each identifying component can be extracted from the multiple vehicle images based on the bounding box. Then, the vehicle detection model can perform multiple recognitions on the same attribute information of the target vehicle (such as vehicle color information, license plate information, or vehicle logo information) (recognizing once for each sub-image). By combining the results of multiple recognitions, it is possible to avoid single recognition failures and improve recognition accuracy.

[0076] Therefore, by tracking the target vehicle using the DeepSORT algorithm, sub-images of the target vehicle and sub-images of each identification component can be extracted from multiple vehicle images, thereby improving recognition accuracy.

[0077] Optionally, according to a vehicle information determination method provided by the present invention, after determining the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, the method further includes:

[0078] The first pixel coordinate position of the target vehicle is determined based on the first bounding box; the second pixel coordinate position of the target vehicle is determined based on the second bounding box; the third pixel coordinate position of the target vehicle is determined based on the third bounding box; and the fourth pixel coordinate position of the target vehicle is determined based on the fourth bounding box.

[0079] Based on camera parameters, the first pixel coordinate position, and the second pixel coordinate position, the first world coordinate position of the target vehicle is determined; and based on the camera parameters, the third pixel coordinate position, and the fourth pixel coordinate position, the second world coordinate position of the target vehicle is determined, wherein the camera parameters are parameters determined by calibrating the binocular camera;

[0080] The speed of the target vehicle is determined based on the first world coordinate position, the second world coordinate position, and the acquisition period.

[0081] Specifically, after determining the first bounding box, second bounding box, third bounding box, and fourth bounding box of the target vehicle, the first pixel coordinate position, second pixel coordinate position, third pixel coordinate position, and fourth pixel coordinate position of the target vehicle can be determined. Then, based on the camera parameters, the first pixel coordinate position, and the second pixel coordinate position, the coordinates of the target vehicle in the pixel coordinate system can be converted into the coordinates in the world coordinate system of the corresponding camera field of view coverage area through the perspective projection matrix. And according to the positional relationship between the field of view coverage areas of each camera (first camera and second camera), the coordinates of the target vehicle in the global world coordinate system at the first moment, that is, the first world coordinate position, can be calculated.

[0082] Specifically, based on camera parameters, the coordinates of the third pixel and the fourth pixel, the coordinates of the target vehicle in the pixel coordinate system can be converted into the coordinates in the world coordinate system of the corresponding camera field of view coverage area through the perspective projection matrix. Furthermore, based on the positional relationship between the field of view coverage areas of each camera (the first camera and the second camera), the coordinates of the target vehicle in the global world coordinate system at the second moment, i.e., the second world coordinate position, can be calculated.

[0083] Specifically, based on the first-world coordinate position and the second-world coordinate position, the running distance of the target vehicle within one acquisition cycle can be determined. Based on the running distance and the acquisition cycle, the speed of the target vehicle can be determined using a speed formula.

[0084] Therefore, by converting the pixel coordinates of the target vehicle to world coordinates, the running distance of the target vehicle within a collection cycle can be determined, and the vehicle speed can be detected.

[0085] Optionally, according to a vehicle information determination method provided by the present invention, the step of extracting sub-images corresponding to the target vehicle and sub-images corresponding to each identifier component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each identifier component includes:

[0086] Based on the first bounding box, a first sub-image of the target vehicle is extracted from the first vehicle image; and based on the second bounding box, a second sub-image of the target vehicle is extracted from the second vehicle image; and based on the third bounding box, a third sub-image of the target vehicle is extracted from the third vehicle image; and based on the fourth bounding box, a fourth sub-image of the target vehicle is extracted from the fourth vehicle image.

[0087] After extracting the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component, the method further includes:

[0088] Based on the camera parameters, the first sub-image, and the second sub-image, a first depth map of the target vehicle is obtained; and based on the camera parameters, the third sub-image, and the fourth sub-image, a second depth map of the target vehicle is obtained.

[0089] Based on the first depth map and the camera position relationship, a first volume is determined; and based on the second depth map and the camera position relationship, a second volume is determined, wherein the camera position relationship is the positional relationship between the first camera and the second camera;

[0090] The volume of the target vehicle is determined based on the first volume and the second volume.

[0091] Specifically, after determining the first, second, third, and fourth bounding boxes of the target vehicle, a first sub-image of the target vehicle can be extracted from the first vehicle image, a second sub-image of the target vehicle can be extracted from the second vehicle image, a third sub-image of the target vehicle can be extracted from the third vehicle image, and a fourth sub-image of the target vehicle can be extracted from the fourth vehicle image based on the bounding boxes of the target vehicle. Then, distortion correction can be performed on the first and second sub-images according to camera parameters. The corrected first and second sub-images are then compared, and the difference between each point in the two sub-images is calculated to obtain a first disparity map. Based on each disparity value in the first disparity map, the corresponding depth value can be obtained according to a preset transformation rule, thus obtaining a first depth map.

[0092] Specifically, distortion correction can be performed on the third and fourth sub-images based on camera parameters. Then, the corrected third and fourth sub-images are compared, and the difference between each point in these two sub-images is calculated to obtain a second disparity map. Based on each disparity value in the second disparity map, the corresponding depth value can be obtained according to a preset conversion rule, thus obtaining a second depth map.

[0093] Specifically, based on the first depth map and the camera position relationship, the first length, first width, and first height of the target vehicle can be determined, and the first volume can be determined using the volume calculation formula. Based on the second depth map and the camera position relationship, the second length, second width, and second height of the target vehicle can be determined, and the second volume can be determined using the volume calculation formula. Furthermore, based on the first and second volumes, the volume of the target vehicle can be determined.

[0094] Optionally, the first height of the target vehicle can be determined based on the relationship between the first depth map and the camera position. The first top envelope of the target vehicle can be determined based on the first depth map. Then, the top envelope can be fitted to determine the first fitted bounding box. Based on the fitted bounding box, the first width and the first height can be determined. Alternatively, the second height of the target vehicle can be determined based on the relationship between the second depth map and the camera position. The second top envelope of the target vehicle can be determined based on the second depth map. Then, the second top envelope can be fitted to determine the second fitted bounding box. Based on the fitted bounding box, the second width and the second height can be determined.

[0095] Optionally, it can be determined whether the first volume and the second volume are within a preset range. If it is determined that both the first volume and the second volume are within the preset range, it indicates that the first volume and the second volume are valid parameters, and then the average value of the first volume and the second volume can be determined as the volume of the target vehicle.

[0096] Optionally, it can be determined whether the first volume and the second volume are within a preset range. If it is determined that the first volume is within the preset range while the second volume is not within the preset range, it indicates that the first volume is a valid parameter, and thus the first volume can be determined as the volume of the target vehicle.

[0097] Optionally, it can be determined whether the first volume and the second volume are within a preset range. If it is determined that the second volume is within the preset range while the first volume is not within the preset range, it indicates that the second volume is a valid parameter, and thus the second volume can be determined as the volume of the target vehicle.

[0098] Understandably, by analyzing multiple sub-images of the target vehicle, multiple predicted volumes of the target vehicle (i.e., the first volume and the second volume) can be obtained. Based on these multiple predicted volumes, the volume of the target vehicle can be determined, which can avoid single detection failures and improve detection accuracy.

[0099] It is understandable that by detecting the speed and volume of each vehicle in the monitored area, the speed and volume of each vehicle in the monitored area can be determined. By statistically analyzing the speed and volume of each vehicle in the monitored area over a period of time, the traffic flow in the monitored area can be predicted.

[0100] Therefore, by analyzing multiple sub-images of the target vehicle, multiple predicted volumes of the target vehicle can be obtained. The volume of the target vehicle can be determined based on multiple predicted volumes, which can avoid single detection failures and improve detection accuracy. By statistically analyzing the speed and volume of each vehicle in the monitored area over a period of time, the traffic flow in the monitored area can be predicted.

[0101] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a license plate detection sub-model, the identification component of the target vehicle includes the target license plate, and the step of determining the identification information of each identification component based on the vehicle detection model and the sub-images corresponding to each identification component includes:

[0102] Based on the license plate detection sub-model and the sub-image corresponding to the target license plate, the recognition information of the target license plate is determined;

[0103] The license plate detection sub-model is obtained by training a Convolutional Recurrent Neural Network (CRNN) model based on the license plate dataset.

[0104] Specifically, in order to identify vehicle license plate information, a CRNN model can be trained in advance based on a license plate dataset to obtain the trained CRNN model, which is a license plate detection sub-model. Then, the sub-image corresponding to the target license plate can be detected based on the license plate detection sub-model to determine the recognition information of the target license plate, that is, the license plate recognition information of the target vehicle.

[0105] Therefore, a vehicle detection model can include a license plate detection sub-model. The license plate detection sub-model can detect the sub-image corresponding to the target license plate and obtain license plate recognition information.

[0106] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a vehicle logo detection sub-model, the identification component of the target vehicle includes a target vehicle logo, and the step of determining the identification information of each identification component based on the vehicle detection model and the sub-images corresponding to each identification component includes:

[0107] Based on the vehicle logo detection sub-model and the sub-image corresponding to the target vehicle logo, the recognition information of the target vehicle logo is determined;

[0108] The vehicle logo detection sub-model is obtained by training a CNN model based on the vehicle logo dataset.

[0109] Specifically, in order to identify vehicle logo information, a CNN model can be pre-trained based on the vehicle logo dataset to obtain the trained CNN model, which is the vehicle logo detection sub-model. Then, the sub-image corresponding to the target vehicle logo can be detected based on the vehicle logo detection sub-model to determine the recognition information of the target vehicle logo, that is, the vehicle logo recognition information of the target vehicle.

[0110] Therefore, a vehicle detection model can include a logo detection sub-model. The logo detection sub-model can detect the sub-image corresponding to the target logo and obtain logo recognition information.

[0111] Optionally, according to a vehicle information determination method provided by the present invention, the vehicle detection model includes a vehicle body color detection sub-model, and the step of determining the vehicle body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle includes:

[0112] Based on the vehicle body color detection sub-model and the sub-image corresponding to the target vehicle, the vehicle body color of the target vehicle is determined;

[0113] The vehicle color detection sub-model is obtained by training a MobileNet model based on the vehicle color dataset.

[0114] Specifically, in order to identify the color of a vehicle's body, a MobileNet model can be trained in advance based on a vehicle color dataset to obtain the trained MobileNet model, which is the vehicle color detection sub-model. Then, the sub-image corresponding to the target vehicle can be detected based on the vehicle color detection sub-model to determine the color of the target vehicle's body.

[0115] Therefore, a vehicle detection model can include a vehicle color detection sub-model. The vehicle color detection sub-model can detect the sub-image corresponding to the target vehicle and obtain the vehicle color.

[0116] The vehicle information determination method provided by this invention detects vehicle image data reported by an image acquisition device using a target detection algorithm. This determines the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle. Based on the bounding boxes, sub-images corresponding to the target vehicle and each identification component can be extracted from the vehicle image data. Furthermore, a vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component. This method can determine the vehicle body color of the target vehicle and the identification information of each identification component, thereby improving the utilization rate of vehicle image data reported by the image acquisition device.

[0117] The vehicle information determination device provided by the present invention is described below. The vehicle information determination device described below and the vehicle information determination method described above can be referred to in correspondence.

[0118] Figure 2 This is a schematic diagram of the vehicle information determination device provided by the present invention, as shown below. Figure 2 As shown, the device includes: a receiving module 201, a first determining module 202, an extracting module 203, and a second determining module 204. Wherein:

[0119] The receiving module 201 is used to receive vehicle image data reported by the image acquisition device;

[0120] The first determining module 202 is used to determine the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, wherein the target vehicle is any vehicle within the monitoring area.

[0121] Extraction module 203 is used to extract sub-images corresponding to the target vehicle and sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component;

[0122] The second determining module 204 is used to determine the body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle, and to determine the identification information of each identification component based on the vehicle detection model and the sub-image corresponding to each identification component.

[0123] The vehicle information determination device provided by this invention detects vehicle image data reported by an image acquisition device using a target detection algorithm. This allows for the determination of the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle. Based on these bounding boxes, sub-images corresponding to the target vehicle and each identification component can be extracted from the vehicle image data. Furthermore, a vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component. This enables the determination of the vehicle body color and the identification information of each identification component, thereby improving the utilization rate of the vehicle image data reported by the image acquisition device.

[0124] Optionally, the vehicle image data includes a first vehicle image, a second vehicle image, a third vehicle image, and a fourth vehicle image. The first vehicle image is an image captured by a first camera at a first moment, the second vehicle image is an image captured by a second camera at the first moment, the third vehicle image is an image captured by the first camera at a second moment, and the fourth vehicle image is an image captured by the second camera at the second moment. The second moment is a moment after the first moment, and there is an interval of one acquisition cycle between the first moment and the second moment. The image acquisition device includes a binocular camera, which includes a first camera and a second camera.

[0125] The first determining module is specifically used for:

[0126] Based on the target detection algorithm, the first bounding box and the second bounding box corresponding to the target vehicle in the first vehicle image and the second vehicle image, respectively, are determined; and based on the target detection algorithm, the bounding box of each first vehicle in the third vehicle image is determined; and based on the target detection algorithm, the bounding box of each second vehicle in the fourth vehicle image is determined.

[0127] Based on the DeepSORT algorithm and the first bounding box, the third bounding box of the target vehicle is determined in the bounding boxes of each first vehicle; and based on the DeepSORT algorithm and the second bounding box, the fourth bounding box of the target vehicle is determined in the bounding boxes of each second vehicle.

[0128] Based on the target detection algorithm, the first bounding box, the second bounding box, the third bounding box, and the fourth bounding box, the fifth bounding box of each identification component of the target vehicle in the first vehicle image, the sixth bounding box of each identification component of the target vehicle in the second vehicle image, the seventh bounding box of each identification component of the target vehicle in the third vehicle image, and the eighth bounding box of each identification component of the target vehicle in the fourth vehicle image are determined.

[0129] Optionally, the device further includes a third determining module, which, after determining the bounding box of the target vehicle and the bounding boxes of each identifying component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, is used to:

[0130] The first pixel coordinate position of the target vehicle is determined based on the first bounding box; the second pixel coordinate position of the target vehicle is determined based on the second bounding box; the third pixel coordinate position of the target vehicle is determined based on the third bounding box; and the fourth pixel coordinate position of the target vehicle is determined based on the fourth bounding box.

[0131] Based on camera parameters, the first pixel coordinate position, and the second pixel coordinate position, the first world coordinate position of the target vehicle is determined; and based on the camera parameters, the third pixel coordinate position, and the fourth pixel coordinate position, the second world coordinate position of the target vehicle is determined, wherein the camera parameters are parameters determined by calibrating the binocular camera;

[0132] The speed of the target vehicle is determined based on the first world coordinate position, the second world coordinate position, and the acquisition period.

[0133] Optionally, the extraction module is specifically used for:

[0134] Based on the first bounding box, a first sub-image of the target vehicle is extracted from the first vehicle image; and based on the second bounding box, a second sub-image of the target vehicle is extracted from the second vehicle image; and based on the third bounding box, a third sub-image of the target vehicle is extracted from the third vehicle image; and based on the fourth bounding box, a fourth sub-image of the target vehicle is extracted from the fourth vehicle image.

[0135] The device further includes a fourth determining module, which, after extracting the sub-images corresponding to the target vehicle and the sub-images corresponding to each identifying component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each identifying component, is used to:

[0136] Based on the camera parameters, the first sub-image, and the second sub-image, a first depth map of the target vehicle is obtained; and based on the camera parameters, the third sub-image, and the fourth sub-image, a second depth map of the target vehicle is obtained.

[0137] Based on the first depth map and the camera position relationship, a first volume is determined; and based on the second depth map and the camera position relationship, a second volume is determined, wherein the camera position relationship is the positional relationship between the first camera and the second camera;

[0138] The volume of the target vehicle is determined based on the first volume and the second volume.

[0139] Optionally, the vehicle detection model includes a license plate detection sub-model, the identification component of the target vehicle includes the target license plate, and the second determining module is specifically used for:

[0140] Based on the license plate detection sub-model and the sub-image corresponding to the target license plate, the recognition information of the target license plate is determined;

[0141] The license plate detection sub-model is obtained by training a CRNN model based on the license plate dataset.

[0142] Optionally, the vehicle detection model includes a vehicle logo detection sub-model, the identification component of the target vehicle includes the target vehicle logo, and the second determining module is specifically used for:

[0143] Based on the vehicle logo detection sub-model and the sub-image corresponding to the target vehicle logo, the recognition information of the target vehicle logo is determined;

[0144] The vehicle logo detection sub-model is obtained by training a CNN model based on the vehicle logo dataset.

[0145] Optionally, the vehicle detection model includes a vehicle body color detection sub-model, and the second determining module is specifically used for:

[0146] Based on the vehicle body color detection sub-model and the sub-image corresponding to the target vehicle, the vehicle body color of the target vehicle is determined;

[0147] The vehicle color detection sub-model is obtained by training a MobileNet model based on the vehicle color dataset.

[0148] The vehicle information determination device provided by this invention detects vehicle image data reported by an image acquisition device using a target detection algorithm. This allows for the determination of the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle. Based on these bounding boxes, sub-images corresponding to the target vehicle and each identification component can be extracted from the vehicle image data. Furthermore, a vehicle detection model can be used to detect the sub-images corresponding to the target vehicle and the sub-images corresponding to each identification component. This enables the determination of the vehicle body color and the identification information of each identification component, thereby improving the utilization rate of the vehicle image data reported by the image acquisition device.

[0149] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 3 As shown, the electronic device may include a processor 310, a communications interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communications interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute a vehicle information determination method, such as the method including:

[0150] Receive vehicle image data reported by the image acquisition device;

[0151] Based on the vehicle image data and target detection algorithm, the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle are determined in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area.

[0152] Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data;

[0153] Based on the vehicle detection model and the sub-image corresponding to the target vehicle, the body color of the target vehicle is determined, and based on the vehicle detection model and the sub-image corresponding to each identification component, the identification information of each identification component is determined.

[0154] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer is able to execute the vehicle information determination method provided by the above methods, for example, the method comprising:

[0156] Receive vehicle image data reported by the image acquisition device;

[0157] Based on the vehicle image data and target detection algorithm, the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle are determined in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area.

[0158] Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data;

[0159] Based on the vehicle detection model and the sub-image corresponding to the target vehicle, the body color of the target vehicle is determined, and based on the vehicle detection model and the sub-image corresponding to each identification component, the identification information of each identification component is determined.

[0160] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle information determination method provided by the methods described above, for example, the method includes:

[0161] Receive vehicle image data reported by the image acquisition device;

[0162] Based on the vehicle image data and target detection algorithm, the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle are determined in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area.

[0163] Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data;

[0164] Based on the vehicle detection model and the sub-image corresponding to the target vehicle, the body color of the target vehicle is determined, and based on the vehicle detection model and the sub-image corresponding to each identification component, the identification information of each identification component is determined.

[0165] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0166] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining vehicle information, characterized in that, include: The system receives vehicle image data reported by an image acquisition device. The vehicle image data includes a sequence of images captured by a camera at multiple consecutive moments. The image sequence includes at least a first image sequence captured by a first camera and a second image sequence captured by a second camera. Based on the vehicle image data and the target detection algorithm, the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle are determined in the pixel coordinate system. The target vehicle is any vehicle within the monitoring area. This includes: determining, based on the target detection algorithm, a first initial bounding box and a second initial bounding box of the target vehicle at the initial time of the first image sequence and the second image sequence, respectively; determining, based on the target tracking algorithm, a first tracking bounding box of the target vehicle at subsequent time points based on the first initial bounding box, and a second tracking bounding box of the target vehicle at subsequent time points based on the second initial bounding box; and determining, based on the bounding boxes at each time point, the bounding boxes of each identification component of the target vehicle at each time point. Based on the bounding box of the target vehicle and the bounding boxes of each marker component, extract the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data; Based on the vehicle detection model and the sub-image corresponding to the target vehicle, the body color of the target vehicle is determined, and based on the vehicle detection model and the sub-images corresponding to each identification component, the identification information of each identification component is determined, including: performing multiple identifications on multiple sub-images of the target vehicle using the vehicle detection model, and combining the results of multiple identifications to determine the body color of the target vehicle; and performing multiple identifications on multiple sub-images of each identification component using the vehicle detection model, and combining the results of multiple identifications to determine the identification information of each identification component.

2. The vehicle information determination method according to claim 1, characterized in that: The consecutive multiple moments include a first moment and a second moment with an interval of one acquisition cycle; The first image sequence includes a first vehicle image captured by the first camera at the first moment and a third vehicle image captured at the second moment; The second image sequence includes a second vehicle image captured by the second camera at the first moment and a fourth vehicle image captured at the second moment; The target tracking algorithm is the DeepSORT algorithm; The first initial bounding box is the first bounding box corresponding to the target vehicle in the first vehicle image, and the second initial bounding box is the second bounding box corresponding to the target vehicle in the second vehicle image; The step of independently tracking and determining the first tracking bounding box of the target vehicle at subsequent time steps based on the first initial bounding box, and independently tracking and determining the second tracking bounding box of the target vehicle at subsequent time steps based on the second initial bounding box, includes: Based on the DeepSORT algorithm and the first bounding box, the third bounding box of the target vehicle is determined in the bounding boxes of each first vehicle in the third vehicle image, and is used as the first tracking bounding box. And based on the DeepSORT algorithm and the second bounding box, the fourth bounding box of the target vehicle is determined in the bounding boxes of each second vehicle in the fourth vehicle image, as the second tracking bounding box; The determination of the bounding boxes of each identifier component of the target vehicle at each time step based on the bounding boxes at each time step includes: Based on the target detection algorithm, the first bounding box, the second bounding box, the third bounding box, and the fourth bounding box, the fifth bounding box of each identification component of the target vehicle in the first vehicle image, the sixth bounding box of each identification component of the target vehicle in the second vehicle image, the seventh bounding box of each identification component of the target vehicle in the third vehicle image, and the eighth bounding box of each identification component of the target vehicle in the fourth vehicle image are determined.

3. The vehicle information determination method according to claim 2, characterized in that, After determining the bounding box of the target vehicle and the bounding boxes of each identifying component of the target vehicle in the pixel coordinate system based on the vehicle image data and the target detection algorithm, the method further includes: The first pixel coordinate position of the target vehicle is determined based on the first bounding box; the second pixel coordinate position of the target vehicle is determined based on the second bounding box; the third pixel coordinate position of the target vehicle is determined based on the third bounding box; and the fourth pixel coordinate position of the target vehicle is determined based on the fourth bounding box. Based on camera parameters, the first pixel coordinate position, and the second pixel coordinate position, the first world coordinate position of the target vehicle is determined; and based on the camera parameters, the third pixel coordinate position, and the fourth pixel coordinate position, the second world coordinate position of the target vehicle is determined, wherein the camera parameters are parameters determined by calibrating the camera; The speed of the target vehicle is determined based on the first world coordinate position, the second world coordinate position, and the acquisition period.

4. The vehicle information determination method according to claim 3, characterized in that, The step of extracting sub-images corresponding to the target vehicle and sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component includes: Based on the first bounding box, a first sub-image of the target vehicle is extracted from the first vehicle image; and based on the second bounding box, a second sub-image of the target vehicle is extracted from the second vehicle image; and based on the third bounding box, a third sub-image of the target vehicle is extracted from the third vehicle image; and based on the fourth bounding box, a fourth sub-image of the target vehicle is extracted from the fourth vehicle image. After extracting the sub-images corresponding to the target vehicle and the sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component, the method further includes: Based on the camera parameters, the first sub-image, and the second sub-image, a first depth map of the target vehicle is obtained; and based on the camera parameters, the third sub-image, and the fourth sub-image, a second depth map of the target vehicle is obtained. Based on the first depth map and the camera position relationship, a first volume is determined; and based on the second depth map and the camera position relationship, a second volume is determined, wherein the camera position relationship is the positional relationship between the first camera and the second camera; The volume of the target vehicle is determined based on the first volume and the second volume.

5. The vehicle information determination method according to any one of claims 1-4, characterized in that, The vehicle detection model includes a license plate detection sub-model, and the identification component of the target vehicle includes the target license plate. The determination of the recognition information of each identification component based on the vehicle detection model and the corresponding sub-images includes: Based on the license plate detection sub-model and the sub-image corresponding to the target license plate, the recognition information of the target license plate is determined; The license plate detection sub-model is obtained by training a CRNN model based on the license plate dataset.

6. The vehicle information determination method according to any one of claims 1-4, characterized in that, The vehicle detection model includes a vehicle logo detection sub-model, and the identification components of the target vehicle include the target vehicle logo. The determination of the recognition information for each identification component based on the vehicle detection model and the corresponding sub-images includes: Based on the vehicle logo detection sub-model and the sub-image corresponding to the target vehicle logo, the recognition information of the target vehicle logo is determined; The vehicle logo detection sub-model is obtained by training a CNN model based on the vehicle logo dataset.

7. The vehicle information determination method according to any one of claims 1-4, characterized in that, The vehicle detection model includes a vehicle body color detection sub-model. The step of determining the vehicle body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle includes: Based on the vehicle body color detection sub-model and the sub-image corresponding to the target vehicle, the vehicle body color of the target vehicle is determined; The vehicle color detection sub-model is obtained by training a MobileNet model based on the vehicle color dataset.

8. A vehicle information determining device, characterized in that, include: The receiving module is used to receive vehicle image data reported by the image acquisition device. The vehicle image data includes image sequences acquired by the camera at multiple consecutive moments. The image sequences include at least a first image sequence acquired by a first camera and a second image sequence acquired by a second camera. The first determining module is configured to determine the bounding box of the target vehicle and the bounding boxes of each identification component of the target vehicle in a pixel coordinate system based on the vehicle image data and the target detection algorithm, wherein the target vehicle is any vehicle within the monitoring area. The module includes: determining a first initial bounding box and a second initial bounding box of the target vehicle at the initial moments of the first image sequence and the second image sequence, respectively, based on the target detection algorithm; independently tracking and determining a first tracking bounding box of the target vehicle at subsequent moments based on the first initial bounding box, and independently tracking and determining a second tracking bounding box of the target vehicle at subsequent moments based on the second initial bounding box, respectively; and determining the bounding boxes of each identification component of the target vehicle at each moment based on the bounding boxes at each moment. The extraction module is used to extract sub-images corresponding to the target vehicle and sub-images corresponding to each marker component from the vehicle image data based on the bounding box of the target vehicle and the bounding boxes of each marker component. The second determining module is used to determine the body color of the target vehicle based on the vehicle detection model and the sub-image corresponding to the target vehicle, and to determine the identification information of each identification component based on the vehicle detection model and the sub-image corresponding to each identification component, including: performing multiple identifications on multiple sub-images of the target vehicle through the vehicle detection model, and determining the body color of the target vehicle by combining the results of multiple identifications; and performing multiple identifications on multiple sub-images of each identification component through the vehicle detection model, and determining the identification information of each identification component by combining the results of multiple identifications.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the vehicle information determination method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the vehicle information determination method as described in any one of claims 1 to 7.

Images