Label configuration method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present application provide a label configuration method and device, electronic equipment and storage medium. The label configuration method comprises the following steps: obtaining image features of vehicle collection images through a feature extraction model, wherein the feature extraction model is trained based on a self-supervised learning method; determining a matching relationship between the image features and the vehicle collection images; clustering the image features according to attribute values between the image features based on the matching relationship to obtain clustering results of the vehicle collection images; determining a label configuration mode based on the clustering results of the vehicle collection images, wherein the label configuration mode is used to configure labels for the vehicle collection images; and obtaining vehicle collection images with labels configured based on the label configuration mode. Based on the matching relationship, different categories of vehicle collection images are obtained by clustering and analyzing the image features. For vehicle collection images of different categories, corresponding label configuration modes are selected, and the accuracy and efficiency of label configuration are taken into account.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a tag configuration method, apparatus, electronic device, and storage medium. Background Technology

[0002] Currently, there are two main methods for configuring labels on vehicle images: one is to manually assign labels to each image, which is accurate but time-consuming and labor-intensive, and unsustainable in the context of big data; the other is to assign labels to each image based on supervised deep learning networks, but this method cannot fully identify objects in the image, which may affect the accuracy of label configuration.

[0003] Therefore, existing solutions cannot simultaneously ensure both the accuracy and efficiency of label configuration. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this application is to provide a tag configuration method, apparatus, electronic device and storage medium to at least partially solve the above problems.

[0005] According to a first aspect of the embodiments of this application, a label configuration method is provided, specifically including the following steps: obtaining image features of vehicle-captured images through a feature extraction model, wherein the feature extraction model is trained based on a self-supervised learning method; determining the matching relationship between the image features and the vehicle-captured images; clustering based on the attribute values ​​between the image features according to the matching relationship to obtain the clustering results of the vehicle-captured images; determining a label configuration mode based on the clustering results of the vehicle-captured images, wherein the label configuration mode is used to configure labels for the vehicle-captured images; and obtaining vehicle-captured images with labels configured based on the label configuration mode.

[0006] Optionally, clustering is performed based on the attribute values ​​between image features to obtain the clustering results of vehicle-acquired images, where the attribute values ​​include density values; based on the clustering results of vehicle-acquired images, a label configuration mode is determined, including: when the density value is less than a preset density, a first label configuration mode is determined, whereby the first label configuration mode configures labels for vehicle-acquired images based on matching relationships.

[0007] Optionally, clustering is performed based on the attribute values ​​between image features to obtain the clustering results of the vehicle-captured images, where the attribute values ​​include distance values; based on the clustering results of the vehicle-captured images, a label configuration mode is determined, including: when the distance value is greater than a preset distance, a first label configuration mode is determined, whereby the first label configuration mode configures labels for the vehicle-captured images based on matching relationships.

[0008] Optionally, acquiring vehicle images with tags configured based on the tag configuration mode includes: acquiring vehicle images with tags configured in the first tag configuration mode.

[0009] Optionally, acquiring vehicle images with tags configured in the first tag configuration mode includes: determining vehicle images corresponding to image features based on matching relationships; sending vehicle images to the tag configuration terminal; and acquiring vehicle images with tags configured based on the first tag configuration mode from the tag configuration terminal.

[0010] Optionally, determining the label configuration mode based on the clustering results of the vehicle-acquired images further includes: determining a second label configuration mode when the density value is greater than a preset density; the second label configuration mode is used to sample image features and configure labels for the vehicle-acquired images based on matching relationships; obtaining vehicle-acquired images with labels configured based on the label configuration mode includes: obtaining vehicle-acquired images with labels configured in the second label configuration mode.

[0011] Optionally, determining the label configuration mode based on the clustering results of the vehicle-captured images further includes: determining a second label configuration mode when the distance value is less than a preset distance; the second label configuration mode is used to sample image features and configure labels for the vehicle-captured images based on matching relationships; obtaining vehicle-captured images with labels configured based on the label configuration mode includes: obtaining vehicle-captured images with labels configured in the second label configuration mode.

[0012] Optionally, acquiring vehicle images with tags configured in the second tag configuration mode includes: obtaining sampled image features based on image features; determining sampled vehicle images corresponding to the sampled image features based on matching relationships; sending the sampled vehicle images to the tag configuration terminal; and acquiring the sampled vehicle images with tags configured based on the second tag configuration mode from the tag configuration terminal.

[0013] According to a second aspect of the embodiments of this application, a label configuration device is also provided, comprising: a feature extraction module, configured to obtain image features of a vehicle-captured image through a feature extraction model, wherein the feature extraction model is trained based on a self-supervised learning method; a determination module, configured to determine the matching relationship between the image features and the vehicle-captured image; a clustering analysis module, configured to perform clustering based on the attribute values ​​between the image features to obtain a clustering result of the vehicle-captured image; a label configuration module, configured to determine a label configuration mode based on the clustering result of the image, wherein the label configuration mode configures labels for the vehicle-captured image based on the matching relationship; and a data acquisition module, configured to acquire the vehicle-captured image with labels configured based on the label configuration mode.

[0014] According to a third aspect of the embodiments of this application, an electronic device is also provided, including: a processor; and a memory storing a program; wherein the program includes instructions that, when executed by the processor, cause the processor to perform the tag configuration method described in the first aspect.

[0015] According to a fourth aspect of the embodiments of this application, a non-transitory computer-readable storage medium storing computer instructions is also provided, wherein the computer instructions are used to cause a computer to execute the tag configuration method described in the first aspect.

[0016] In the scheme of this application embodiment, image features of vehicle-captured images are obtained through a feature extraction model trained based on a self-supervised learning method; the matching relationship between image features and vehicle-captured images is determined; based on the matching relationship, clustering is performed according to the attribute values ​​between image features to obtain the clustering results of vehicle-captured images; based on the clustering results of vehicle-captured images, a label configuration mode is determined; and vehicle-captured images with labels configured according to the label configuration mode are obtained. Based on the matching relationship between image features and vehicle-captured images, different categories of vehicle-captured images are obtained through cluster analysis of image features. For different categories of vehicle-captured images, the corresponding label configuration mode is selected, balancing the accuracy and efficiency of label configuration. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application 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 only some embodiments recorded in the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a diagram of a vehicle-to-everything (V2X) communication architecture based on an example.

[0019] Figure 2 This is a flowchart of the steps of a tag configuration method provided according to an embodiment of this application.

[0020] Figure 3 This is a flowchart illustrating the steps of a method for acquiring vehicle images with configuration tags according to an embodiment of this application.

[0021] Figure 4 This is a flowchart of another method for acquiring vehicle images with configuration tags according to an embodiment of this application.

[0022] Figure 5 This is a schematic diagram of a tag configuration device provided according to an embodiment of this application.

[0023] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0024] To provide a clearer understanding of the technical features, objectives, and effects of the embodiments of this application, the specific implementation methods of the embodiments of this application will now be described with reference to the accompanying drawings.

[0025] In this document, “illustrative” means “serving as an example, illustration or description”, and any illustration or implementation described herein as “illustrative” should not be construed as a more preferred or advantageous technical solution.

[0026] To keep the drawings concise, only the parts relevant to this application are shown schematically in each drawing, and they do not represent the actual structure of the product. In addition, to make the drawings concise and easy to understand, some parts with the same structure or function are only shown schematically, or only one or more of them are labeled.

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] Figure 1 As shown in the figure, a vehicle-to-everything (V2X) communication architecture diagram illustrates the data communication between vehicle 110 and cloud server 120, such as communication regarding vehicle image acquisition. Specifically, vehicle 110 can wirelessly transmit acquired vehicle images to cloud server 120. Cloud server 120 automatically identifies image features contained in the acquired vehicle images using a supervised deep learning network, assigns corresponding labels to each received vehicle image based on these features, and stores the labeled vehicle images in cloud server 120 for subsequent model development or training. However, in this label configuration method, cloud server 120 uses a supervised deep learning network to identify vehicle images. If the vehicle images transmitted from vehicle 110 to cloud server 120 are not recognizable by the supervised deep learning network, it can easily lead to incorrect label configuration, affecting the accuracy of image label configuration. Furthermore, cloud server 120 may save incorrectly labeled vehicle images for subsequent model training and development, potentially impacting the accuracy of subsequent model training and development.

[0029] Figure 2 This is a flowchart illustrating the steps of a tag configuration method according to an embodiment of this application. Figure 2 As shown, this embodiment mainly includes the following steps:

[0030] Step 201: Obtain image features of the vehicle-collected image through a feature extraction model, which is trained based on a self-supervised learning method.

[0031] In the above implementation, the vehicle images are acquired from the vehicle 110 via wireless communication. The wireless communication can be 3G, 4G, 5G or other communication methods. The vehicle 110 can determine the network quality of the wireless communication network technology and automatically select a wireless communication method with better network quality to upload the vehicle images to the cloud server 120 to ensure the stability and reliability of data transmission. Alternatively, the specific communication method can be manually selected. This application does not limit the specific means of selecting the communication method.

[0032] Step 202: Determine the matching relationship between image features and vehicle-acquired images.

[0033] Specifically, image features can be bound to the corresponding vehicle-captured image based on the image ID (unique identifier) ​​to determine the matching relationship between the image features and the vehicle-captured image.

[0034] Step 203: Based on the matching relationship, clustering is performed according to the attribute values ​​between image features to obtain the clustering results of the vehicle acquisition images.

[0035] Specifically, the attribute values ​​between image features include distance or density values, or other attribute values. Based on the attribute values ​​between image features and the matching relationship between image features and vehicle-captured images, the clustering results of the vehicle-captured images are determined. Specifically, algorithms such as K-means clustering and density clustering can be used to perform clustering analysis on the image features to obtain different categories of vehicle-captured images. Based on the matching relationship, efficient and accurate clustering of vehicle-captured images is achieved by clustering image features.

[0036] Step 204: Determine the label configuration mode based on the clustering results of the vehicle images.

[0037] Specifically, based on the clustering results of the vehicle images, different label configuration modes are configured for different categories of vehicle images. By selecting different label configuration modes based on the image features of different categories, both the accuracy and efficiency of label configuration are balanced.

[0038] Step 205: Obtain vehicle images with tags configured based on the tag configuration mode.

[0039] Specifically, vehicle images with labels configured based on different label configuration modes are acquired and stored to facilitate the development and training of subsequent models.

[0040] In this embodiment, based on the matching relationship between image features and vehicle-captured images, different categories of vehicle-captured images are obtained by performing cluster analysis on the image features. Thus, a corresponding label configuration mode is selected for different categories of vehicle-captured images, taking into account both the accuracy and efficiency of label configuration.

[0041] Furthermore, in this embodiment, a feature extraction model can be built using a self-supervised learning-based backbone network and projector network. The feature extraction model is trained on a cloud server using a dataset consisting of unlabeled vehicle images. After training, the backbone network is used to extract image features from the vehicle images. The backbone network is continuously adjusted based on updates to the dataset.

[0042] Specifically, when implementing step 203 above, taking the K-means clustering algorithm as an example, when executing the clustering algorithm, the center vector of the image features to be processed can be set according to the preset number of labels, the distance of each image feature in the image features to be processed to the center vector can be calculated, and then the clustering categories corresponding to the images of the vehicles to be processed can be divided according to the minimum distance of the image features to the center vector to obtain multiple image features.

[0043] For each image feature in the image to be processed, it can be regarded as a data point in a multi-dimensional space. In the initial clustering, since the number of target categories, such as k, can be determined according to the number of preset labels (k can be a natural number and set according to different needs), that is, the image features to be processed need to be divided into k categories. Therefore, the center vectors of each image feature to be processed can be randomly initialized based on the specified number of target categories. Then, the distance to each selected center vector is calculated for other image features, and then each image feature is classified into the center vector with the closest distance.

[0044] In one specific implementation, during the execution of step 203 above, clustering is performed based on the attribute values ​​between image features to obtain multiple types of image features. The attribute values ​​include density values. When the density value is less than a preset density, a first label configuration mode is determined. The first label configuration mode is used to configure labels for vehicle-collected images.

[0045] Specifically, when the density value of the image features is less than the preset density, the specific mode for configuring the label of the vehicle-collected image is the first label configuration mode. The specific value of the preset density is determined based on the actual situation. This application does not limit the specific value of the preset ratio threshold.

[0046] In one specific implementation method, the above step 205 is specifically implemented as follows, mainly including:

[0047] Acquire vehicle images with tags configured in the first tag configuration mode.

[0048] In the above implementation, the vehicle images with labels configured based on the first label configuration mode are stored to facilitate the development and training of subsequent models.

[0049] like Figure 3 As shown, Figure 3 This is a flowchart illustrating a method for acquiring vehicle images with configured tags according to an embodiment of this application. This embodiment is a specific implementation of acquiring vehicle images with configured tags in a first tag configuration mode, and mainly includes the following steps:

[0050] Step 2511: Send the vehicle-captured images to the tag configuration terminal.

[0051] Specifically, vehicle images are sent to the tag configuration terminal. Staff obtain the specific content of the vehicle images through the tag configuration terminal, perform semantic analysis on the vehicle images based on the specific content, determine the corresponding tags for the vehicle images in the preset tag dictionary, and complete the tag configuration for the vehicle images.

[0052] By performing cluster analysis on image features to obtain multi-class image features, and then manually performing semantic analysis on vehicle-collected images based on the classified image features, the accuracy of label configuration can be improved.

[0053] Step 2512: Obtain vehicle images from the tag configuration terminal based on the first tag configuration mode.

[0054] Specifically, the system retrieves vehicle images with tags configured by staff from the tag configuration terminal and stores these tagged vehicle images for subsequent model updates and development. The model can be any model used in the process of improving intelligent driving.

[0055] In one specific embodiment, step 204 above is implemented as follows, mainly including:

[0056] When the density value between image features is greater than the preset density, a second label configuration mode is determined. The second label configuration mode is used to sample image features and configure labels for vehicle images.

[0057] Specifically, when the density values ​​among multiple types of image features are greater than the preset density, the specific mode for configuring labels on the vehicle-acquired images is the second label configuration mode. The specific value of the preset density is determined based on the actual situation. This application does not limit the specific value of the preset ratio threshold.

[0058] In one specific implementation, step 205 may further include:

[0059] Acquire vehicle images with the tags configured in the second tag configuration mode.

[0060] In the above implementation, the vehicle images with labels configured based on the second label configuration mode are stored to facilitate the development and training of subsequent models.

[0061] like Figure 4 As shown, Figure 4 This is a flowchart illustrating another method for obtaining vehicle images with configured tags according to an embodiment of this application. This embodiment is a specific implementation of obtaining vehicle images with configured tags in a second tag configuration mode, and mainly includes the following steps:

[0062] Step 2521: Obtain sampled vehicle images based on the vehicle images;

[0063] Specifically, the vehicle images collected after clustering are sampled to obtain sampled vehicle images. Regarding the specific implementation method of sampling, a small number of vehicle images can be randomly selected, or the vehicle images can be sampled according to the set rules. This application does not restrict the specific implementation method of sampling.

[0064] Step 2522: Send the collected images of the sampled vehicles to the tag configuration terminal.

[0065] Specifically, the sampled vehicle images are sent to the tag configuration terminal. Staff members obtain the specific content of the sampled vehicle images through the tag configuration terminal, perform semantic analysis based on the specific content of the sampled vehicle images, determine the corresponding tags for the sampled vehicle images in the preset tag thesaurus, and complete the tag configuration for the sampled vehicle images.

[0066] By performing cluster analysis on image features, and obtaining image features of different categories, the image features are then sampled, reducing the workload of manually configuring labels and improving the efficiency of label configuration.

[0067] Step 2523: Obtain vehicle images from the tag configuration terminal based on the second tag configuration mode.

[0068] Specifically, the system obtains the sampled vehicle images with labels configured by the staff from the label configuration terminal, determines the image feature category corresponding to the features of the sampled images, configures the same label for the vehicle images corresponding to the image feature category, and stores the tagged vehicle images for subsequent model updates and development. Here, the model can be any model used in the process of improving intelligent driving.

[0069] By performing cluster analysis on image features, and obtaining image features of different categories, the image features are then sampled, reducing the workload of manual label configuration and improving the efficiency of label configuration. The sampled images are sent to the label configuration terminal, and labels are configured for vehicle images manually, ensuring the accuracy of label configuration and further improving the efficiency of label configuration.

[0070] In another implementation, the attribute values ​​between image features include distance values. When the distance value is greater than a preset distance, a first label configuration mode is determined, which assigns labels to vehicle images based on matching relationships. When the distance value is less than a preset distance, a second label configuration mode is determined, which samples image features and assigns labels to vehicle images based on matching relationships. The specific value of the preset distance is determined based on actual conditions. The specific operation steps and corresponding technical effects of the first and second label configuration modes have been described in detail above in the specific implementation of obtaining vehicle images with labels configured in the first label configuration mode and in a specific implementation of obtaining vehicle images with labels configured in the second label configuration mode, and will not be repeated here.

[0071] Reference Figure 5 , Figure 5 The present invention is a schematic diagram of a tag configuration device 300 provided according to an embodiment of the present application. The tag configuration device specifically includes: a feature extraction module 301, a matching relationship determination module 302, a cluster analysis module 303, a tag configuration module 304, and a data acquisition module 305.

[0072] The feature extraction module 301 is used to obtain image features of the vehicle-collected image through the feature extraction model, which is trained based on a self-supervised learning method.

[0073] The determination module 302 is used to determine the matching relationship between image features and vehicle-acquired images.

[0074] The clustering analysis module 303 is used to perform clustering based on the matching relationship and the attribute values ​​between image features to obtain the clustering results of the vehicle acquisition images.

[0075] The label configuration module 304 is used to determine the label configuration mode based on the clustering results of the vehicle-acquired images. The label configuration mode is used to configure labels for the vehicle-acquired images.

[0076] The data acquisition module 305 is used to acquire vehicle images based on the tag configuration mode.

[0077] In this embodiment, by performing cluster analysis on image features, different classifications of image features are obtained, thereby selecting the corresponding label configuration mode for multiple types of image features, which balances the accuracy and efficiency of label configuration.

[0078] In one specific embodiment, the tag configuration module includes a first tag configuration module. The first tag configuration module is used to determine a first tag configuration mode when the density value between image features is less than a preset density or the distance value between image features is greater than a preset distance. The first tag configuration mode is used to configure tags for images acquired from vehicles.

[0079] In one specific embodiment, the data acquisition module is further configured to acquire vehicle images with tags configured in the first tag configuration mode.

[0080] In one specific embodiment, the data acquisition module specifically includes: an image sending module, used to send vehicle-collected images to a tag configuration terminal; and an image acquisition module, used to acquire vehicle-collected images configured with tags based on a first tag configuration mode from the tag configuration terminal.

[0081] In one specific embodiment, the label configuration module includes a second label configuration module. The second label configuration module is used to determine a second label configuration mode when the density value between image features is greater than a preset density or the distance value between image features is less than a preset distance. The second label configuration mode is used to sample image features and configure labels for vehicle-collected images.

[0082] In one specific embodiment, the data acquisition module is further configured to acquire vehicle images with tags configured in the second tag configuration mode.

[0083] In one specific embodiment, the data acquisition module specifically includes: an image sampling module, used to obtain sampled image features based on image features; an image sending module, further used to send vehicle-collected images to a tag configuration terminal; and an image acquisition module, further used to acquire vehicle-collected images configured with tags based on a second tag configuration mode from the tag configuration terminal.

[0084] The label configuration device 300 of this application embodiment is used to implement other steps in the aforementioned label configuration method embodiment, and has the beneficial effects of the corresponding method step embodiment, which will not be repeated here.

[0085] Exemplary embodiments of this application also provide a non-transitory computer-readable storage medium storing computer instructions, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform methods according to various embodiments of this application.

[0086] Exemplary embodiments of this application also provide a computer program product, including a computer program, wherein, when executed by a computer's processor, the computer program is used to cause the computer to perform methods according to various embodiments of this application.

[0087] Reference Figure 6 The diagram shows a schematic of an electronic device according to another embodiment of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the electronic device.

[0088] like Figure 6 As shown, the electronic device may include: a processor 602, a communications-interface 604, a memory 606 storing a program 610, and a communications bus 608.

[0089] The processor 602, communication interface 604, and memory 606 communicate with each other via communication bus 608.

[0090] Communication interface 604 is used for communication with other electronic devices or servers.

[0091] The processor 602 is used to execute programs, specifically the relevant steps in the above method embodiments.

[0092] Specifically, the program may include program code, which includes computer operation instructions.

[0093] Processor 602 may be a CPU, an Application-Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The smart device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0094] Memory 606 is used to store programs. Memory may include high-speed RAM and may also include non-volatile memory, such as at least one disk drive.

[0095] The program can be used to cause the processor to execute the label configuration method described above.

[0096] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0097] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0098] As used in this application, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0099] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0100] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0101] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0102] This application also provides a computer storage medium storing a computer executable program, which is run to implement any of the tag configuration methods described in the above embodiments.

[0103] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0104] The above descriptions are merely illustrative embodiments of this application and are not intended to limit the scope of the embodiments of this application. Any equivalent changes, modifications, and combinations made by those skilled in the art without departing from the concept and principles of the embodiments of this application should fall within the protection scope of the embodiments of this application.

Claims

1. A label configuration method, characterized in that, include: Image features of vehicle-captured images are obtained through a feature extraction model, which is trained based on a self-supervised learning method. Determine the matching relationship between the image features and the vehicle-acquired image; Based on the matching relationship, clustering is performed according to the attribute values ​​between the image features to obtain the clustering result of the vehicle-collected images. The attribute values ​​include density values ​​or distance values. Based on the clustering results of the vehicle-captured images, a label configuration mode is determined. This label configuration mode is used to configure labels for the vehicle-captured images. If the attribute value is the density value, then if the density value is less than a preset density based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be a first label configuration mode. The first label configuration mode is used to configure labels for the vehicle-captured images. If the density value is greater than the preset density based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be a second label configuration mode. The second label configuration mode is used to sample the vehicle-captured images and configure labels for them. If the attribute value is the distance value, then if the distance value is greater than a preset distance based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be the first label configuration mode. If the distance value is less than the preset distance based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be the second label configuration mode. Acquire the vehicle images based on the tags configured according to the tag configuration mode.

2. The label configuration method according to claim 1, characterized in that, The step of acquiring the vehicle-collected image based on the tag configuration mode includes: Acquire the vehicle image with the tag configured in the first tag configuration mode.

3. The label configuration method according to claim 2, characterized in that, The step of acquiring the vehicle image with the tag configured in the first tag configuration mode includes: Send the captured vehicle images to the tag configuration terminal; The vehicle image is acquired from the tag configuration terminal, with the tag configured based on the first tag configuration mode.

4. The label configuration method according to claim 1, characterized in that, The step of acquiring the vehicle-collected image based on the tag configuration mode includes: Acquire the vehicle image with the tag configured in the second tag configuration mode.

5. The label configuration method according to claim 4, characterized in that, The step of acquiring the vehicle image with the tag configured in the second tag configuration mode includes: Based on the vehicle acquisition images, sampled vehicle acquisition images are obtained; Send the images of the sampled vehicles to the tag configuration terminal; The sampled vehicle images are obtained from the label configuration terminal based on the second label configuration mode.

6. A label configuration device, characterized in that, include: The feature extraction module is used to obtain image features of vehicle-collected images through a feature extraction model, which is trained based on a self-supervised learning method. A determining module is used to determine the matching relationship between the image features and the vehicle-acquired image; The clustering analysis module is used to cluster the images based on the matching relationship and the attribute values ​​between the image features to obtain the clustering results of the vehicle-collected images. The attribute values ​​include density values ​​or distance values. A label configuration module is used to determine a label configuration mode based on the clustering results of the vehicle-captured images. The label configuration mode is to configure labels for the vehicle-captured images. If the attribute value is the density value, then if the density value is less than a preset density based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be a first label configuration mode. The first label configuration mode is used to configure labels for the vehicle-captured images. If the density value is greater than the preset density based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be a second label configuration mode. The second label configuration mode is used to sample the vehicle-captured images and configure labels for them. If the attribute value is the distance value, then if the distance value is greater than a preset distance based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be the first label configuration mode. If the distance value is less than the preset distance based on the clustering results of the vehicle-captured images, the label configuration mode is determined to be the second label configuration mode. The data acquisition module is used to acquire the vehicle images based on the tags configured according to the tag configuration mode.

7. An electronic device, comprising: processor; as well as Memory for stored programs; The program includes instructions that, when executed by the processor, cause the processor to perform the method according to any one of claims 1-5.

8. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-5.

Images