Data processing method, device, apparatus, and computer storage medium

Abstract

This application discloses a data processing method, apparatus, device, and computer storage medium. It can acquire multiple scene data collected by camera sensors and radar sensors during vehicle operation. Then, it performs target detection on the scene data using a point cloud detection model and an image detection model, obtaining a first target detection result and a second target detection result, thereby determining the value score of each scene data. Finally, based on the n scene data with higher value scores, it determines the training data for the target detection model. According to the embodiments of this application, the training data for the target detection model can be determined based on several scene data with higher value scores. This value score is determined based on the results of target detection on the scene data using multiple target detection models. Therefore, by using scene data collected by multiple sensors, the scene data with higher value scores can be determined more accurately and efficiently, thereby determining the training samples more accurately and efficiently.

Description

Technical Field

[0001] This application belongs to the field of data processing, and in particular relates to a data processing method. Background Technology

[0002] With the development of computer and information technology, autonomous driving equipment is becoming increasingly sophisticated. During operation, autonomous driving equipment typically collects data about its surrounding environment using sensors such as cameras and radar. It then uses target detection models to identify obstacles from this data, thereby reducing the likelihood of collisions. Therefore, the accuracy of the target detection model in detecting obstacles is crucial to the driving safety of autonomous driving equipment.

[0003] Typically, when training an object detection model, in order to improve efficiency, users will train the model using some hard-to-identify samples and some samples with long-tail effects.

[0004] However, when identifying difficult-to-identify samples or samples with long-tail effects, users often rely solely on data collected by a single sensor, neglecting the correlation between data collected by different sensors, resulting in low accuracy and efficiency. Summary of the Invention

[0005] This application provides a data processing method, apparatus, device, and computer storage medium that can improve the efficiency of determining training samples for a target detection model.

[0006] On one hand, embodiments of this application provide a data processing method, the method comprising:

[0007] The system acquires multiple scene data collected during vehicle operation, the scene data being collected by camera sensors and radar sensors configured in the vehicle.

[0008] For each scene data, a point cloud detection model is used to perform target detection on the scene data to obtain a first target detection result; an image detection model is used to perform target detection on the scene data to obtain a second target detection result.

[0009] Based on the first target detection result and the second target detection result, determine the value score of the scene data;

[0010] The data from multiple scenarios are sorted in descending order of value scores to obtain the first sequence.

[0011] The training data for the target detection model is determined based on the first n scene data in the first sequence.

[0012] Optionally, determining the value score of the scene data based on the first target detection result and the second target detection result specifically includes:

[0013] The scene data is subjected to target detection by a target detection model to obtain a third target detection result. The input data of the target detection model is radar point cloud or camera image.

[0014] Based on the detection results of the third target, a detection accuracy score for the scene data is determined;

[0015] Based on the detection results of the first target, the second target, and the third target, a detection difficulty score is determined.

[0016] Based on the detection accuracy score and the detection difficulty score, the value score of the scene data is determined using the following formula:

[0017] S = c1S confidence +c2S rarity

[0018] Where S is the value score, and c1 and c2 are constants, S confidence S is used to score the accuracy of the detection. rarity The difficulty of the detection is scored.

[0019] Optionally, the third target detection result includes a bounding box and the confidence level corresponding to the bounding box. The step of determining the detection accuracy score of the scene data based on the third target detection result specifically includes:

[0020] The number of target boxes contained in the third target detection result is determined as the number of target boxes;

[0021] Based on the number of bounding boxes and the confidence level of each bounding box, the detection accuracy score for this scene data is determined using the following formula:

[0022]

[0023] Among them, S confidence To score the accuracy of the detection, n obj ss represents the number of target boxes. camera The confidence level is the value corresponding to the target bounding box.

[0024] Optionally, determining the detection difficulty score based on the first target detection result, the second target detection result, and the third target detection result specifically includes:

[0025] When the input data of the target detection model is a camera image, the second target detection result is determined as the comparison target detection result;

[0026] When the input data of the target detection model is radar point cloud, the first target detection result is determined as the comparison target detection result;

[0027] Based on the first target detection result and the second target detection result, a first detection difficulty score is determined;

[0028] Based on the third target detection result and the comparison target detection result, a second detection difficulty score is determined;

[0029] Based on the first detection difficulty score and the second detection difficulty score, the detection difficulty score of the scene data is determined using the following formula:

[0030] S rarity =αLS rarity +βSS rarity

[0031] Among them, S rarity The detection difficulty is scored, where α and β are constants, and LS rarity The first test difficulty score is SS. rarity The second test difficulty score is given.

[0032] Optionally, the first target detection result includes a target bounding box and the confidence level corresponding to the target bounding box; the second target detection result includes a target bounding box and the confidence level corresponding to the target bounding box; and determining a first detection difficulty score based on the first target detection result and the second target detection result specifically includes:

[0033] The number of target boxes contained in the first target detection result is determined as the number of target boxes;

[0034] From the target boxes included in the first target detection result, the target boxes that match any target box included in the second target detection result are determined as matching target boxes;

[0035] The number of matching target boxes, the confidence level of the matching target boxes in the first target detection result, and the matching confidence level of the target boxes that match the matching target boxes in the second target detection result are determined.

[0036] The matching value is determined by summing the confidence score of the matching target box with the matching confidence score.

[0037] Based on the number of matching bounding boxes, the number of bounding boxes, and the matching value, the first detection difficulty score for this scene data is determined using the following formula:

[0038]

[0039] Among them, LS rarity The difficulty score for the first detection is nll match lsmatch represents the number of matched target boxes. i Let n be the matching value of the i-th target box. lidar The number of target boxes.

[0040] Optionally, the third target detection result includes a bounding box and the confidence level corresponding to the bounding box; the comparison target detection result includes a bounding box and the confidence level corresponding to the bounding box; and determining the second detection difficulty score based on the third target detection result and the comparison target detection result specifically includes:

[0041] The number of target boxes contained in the comparison target detection results is determined as the number of target boxes;

[0042] From the target boxes included in the comparison target detection results, the target boxes that match any target box included in the third target detection results are determined as similar target boxes;

[0043] The number of similar target boxes, the confidence level of the similar target boxes in the comparison target detection results, and the similarity confidence level of the target boxes that match the similar target boxes in the third target detection results are determined.

[0044] The similarity value is determined by summing the confidence scores of the similar bounding boxes and the similarity scores.

[0045] Based on the number of similar bounding boxes, the total number of bounding boxes, and the similarity value, the first detection difficulty score for this scene data is determined using the following formula:

[0046]

[0047] Among them, SS rarity The second detection difficulty score is given by nls. match ssmatch represents the number of similar bounding boxes. i Let nl be the similarity value of the i-th similar bounding box. camera The number of target boxes.

[0048] Optionally, determining the training data for the target detection model based on the first n scene data in the first sequence specifically includes:

[0049] The first n scene data in the first sequence are labeled to obtain multiple training data;

[0050] The target detection model is trained using the aforementioned training data.

[0051] Optionally, after determining the training data for the target detection model based on the first n scene data in the first sequence, the method further includes:

[0052] Obtain the collection time identifier of the training data;

[0053] The training data is sorted according to the order of the collection time identifiers to obtain a second sequence;

[0054] Each pair of adjacent training data in the second sequence is determined as a training sample group;

[0055] The training data that ranks first among the two training data contained in the training sample group is determined as the first training data, and the training data that ranks last is determined as the second training data.

[0056] For each training sample group in the second sequence, when it is determined that the structural similarity between two training data in the training sample group is greater than a preset similarity threshold, the second training data in the training sample group is deleted until the structural similarity between two training data in any training sample group in the second sequence is less than or equal to the similarity threshold.

[0057] On the other hand, embodiments of this application provide a data processing apparatus, the apparatus comprising:

[0058] The acquisition unit is used to acquire multiple scene data collected by the vehicle during driving, the scene data being collected by the camera sensor and radar sensor configured in the vehicle;

[0059] The detection unit is used to perform target detection on each scene data using a point cloud detection model to obtain a first target detection result; and to perform target detection on the scene data using an image detection model to obtain a second target detection result.

[0060] The scoring unit is used to determine the value score of the scene data based on the first target detection result and the second target detection result;

[0061] The sorting unit is used to sort multiple scenario data in descending order of value score to obtain the first sequence;

[0062] The determining unit is used to determine the training data of the target detection model based on the first n scene data in the first sequence.

[0063] Furthermore, embodiments of this application provide a data processing apparatus, the apparatus comprising:

[0064] Processor and memory storing computer program instructions;

[0065] When the processor executes the computer program instructions, it implements the data processing method as described in one aspect above.

[0066] In another aspect, embodiments of this application provide a computer storage medium storing computer program instructions, which, when executed by a processor, implement the data processing method as described in any of the preceding aspects.

[0067] In another aspect, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to implement the data processing method as described in any of the preceding aspects.

[0068] The data processing method, apparatus, device, and computer storage medium of this application embodiment can determine the training data of the target detection model based on a number of scene data with high value scores. The value scores are determined based on a first target detection result and a second target detection result. The first target detection result is obtained by performing target detection on the scene data using a point cloud detection model, and the second target detection result is obtained by performing target detection on the scene data using an image detection model. Therefore, by using scene data collected from multiple sensors, scene data with high value scores can be determined more accurately and efficiently, thereby determining training samples more accurately and efficiently. Attached Figure Description

[0069] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1 This is a schematic flowchart of a data processing method provided in one embodiment of this application;

[0071] Figure 2 This is a flowchart illustrating another data processing method provided in one embodiment of this application;

[0072] Figure 3 This is a schematic diagram of the structure of a data processing apparatus provided in one embodiment of this application;

[0073] Figure 4This is a schematic diagram of the structure of a data processing device provided in one embodiment of this application. Detailed Implementation

[0074] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0075] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0076] As described in the background section, the accuracy of perception in autonomous driving devices is crucial, relying on massive amounts of training data. However, much of this data is repetitive, low-quality, or invalid. Directly labeling this data would waste resources and reduce efficiency. Therefore, when training object detection models, users often use samples that are difficult to identify and samples with long-tail effects.

[0077] However, when identifying difficult-to-identify samples or samples with long-tail effects, users often rely solely on data collected by a single sensor, neglecting the correlation between data collected by different sensors, resulting in low accuracy and efficiency.

[0078] To address the problems of the prior art, embodiments of this application provide a data processing method, apparatus, device, and computer storage medium. The data processing method provided in this application embodiment will be described first below.

[0079] Figure 1 A flowchart illustrating a data processing method provided in one embodiment of this application is shown. Figure 1As shown, the data processing method provided in this application embodiment includes the following steps: S101 to S105.

[0080] S101: Acquire multiple scene data collected during vehicle operation. The scene data is collected by the camera sensors and radar sensors configured in the vehicle.

[0081] In one or more embodiments of this application, scene data can be perception data of the current scene collected by the vehicle's camera sensors and radar sensors during vehicle operation. The data processing method can be determined by an electronic device. Of course, this electronic device can be a vehicle control center, mobile phone, tablet computer, server, etc. This application does not limit the specific device used and can set it as needed. The scene data includes image data and point cloud data. The image data and the point cloud data correspond. That is, if a first obstacle exists at a first location in the image data, an obstacle identical to the first obstacle also exists at the same location in the point cloud data.

[0082] Using the above method, multiple scene data collected during vehicle operation can be obtained.

[0083] S102: For each scene data, the point cloud detection model is used to perform target detection on the scene data to obtain the first target detection result; the image detection model is used to perform target detection on the scene data to obtain the second target detection result.

[0084] In one or more embodiments of this application, the point cloud detection model is obtained by training a point cloud 3D detection network based on a labeled multimodal dataset. The first object detection result is the object detection result obtained by performing object detection on the scene data using the point cloud detection model. The image detection model is obtained by training a visual 3D detection network based on a labeled multimodal dataset. The second object detection result is the detection result obtained by performing object detection on the scene data using the image detection model.

[0085] As an example, the training process for the point cloud detection model and the image detection model described above can be as follows: A large amount of training sample data is acquired to obtain a preliminary training model, namely the point cloud detection model and the image detection model. Then, the preliminary training model is used to filter training data through steps S101 to S105, and the preliminary training model is trained using the filtered training data. The above steps are repeated iteratively until the preliminary training model converges.

[0086] Using the above method, the first and second target detection results of the scene data can be obtained, which can be used to determine the value score of the scene data.

[0087] S103: Determine the value score of the scene data based on the first target detection result and the second target detection result.

[0088] In one or more embodiments of this application, the value score of the scene data can be calculated based on the first target detection result and the second target detection result.

[0089] Using the above method, the value score of the scene data can be determined, which can then be used to rank the scene data.

[0090] S104: Sort the data from multiple scenarios in descending order of value scores to obtain the first sequence.

[0091] In one or more embodiments of this application, the scene data corresponding to the value scores are sorted from largest to smallest according to the value scores. The first sequence can be a sorted sequence of multiple scene data.

[0092] As an example, if scenario data A has a value score of 90, scenario data B has a value score of 97, and scenario data C has a value score of 80, then the first sequence is: scenario data B, scenario data A, and scenario data C.

[0093] Using the above method, the first sequence of scene data can be obtained, which can be used to determine the training data for the target detection model.

[0094] S105: Determine the training data for the object detection model based on the first n scene data in the first sequence.

[0095] In one or more embodiments of this application, the first n scene data in the first sequence can be directly used as the training data of the object detection model, or the first n scene data in the first sequence can be processed, for example, by filtering out duplicate data in the scene data, to obtain the training data of the object detection model. This is not limited here.

[0096] The above describes the specific implementation of the data processing method provided in this application. It is evident that in the above embodiments, the training data for the target detection model can be determined based on several scene data points with high value scores. These value scores are determined based on a first target detection result and a second target detection result. The first target detection result is obtained by performing target detection on the scene data using a point cloud detection model, and the second target detection result is obtained by performing target detection on the scene data using an image detection model. Therefore, by using scene data collected from multiple sensors, scene data with high value scores can be determined more accurately and efficiently, thereby determining training samples more accurately and efficiently.

[0097] To more accurately and efficiently determine the training samples for the object detection model, as another implementation of this application, an implementation of S103 is also provided, which includes:

[0098] The scene data is used to detect targets using a target detection model to obtain a third target detection result. The input data for the target detection model is radar point cloud or camera image.

[0099] Based on the detection results of the third target, a detection accuracy score for the scene data is determined.

[0100] The detection difficulty score is determined based on the detection results of the first target, the second target, and the third target.

[0101] Based on the detection accuracy score and the detection difficulty score, the value score of the data in this scenario is determined using the following formula:

[0102] S = c1S confidence +c2S rarity (1)

[0103] Where S is the value rating, and c1 and c2 are constants, S confidence To score the accuracy of the detection, S rarity To assess the difficulty level of the test.

[0104] In the above implementation, the target detection model can be a detection model that takes radar point cloud data or camera image data as input and outputs a third target detection result. When the target detection model is an image target detection model, the input is a camera image; when the target detection model is a radar target detection model, the input is radar point cloud data. The third target detection result can be the detection result of scene data, such as the number of target boxes in the current scene.

[0105] In some embodiments, due to inherent errors in the object detection model's detection of scene data—for example, the identification of an obstacle might result in both correct and incorrect detections—the third object detection result output by the object detection model is statistically analyzed and calculated as a score. This score is the detection accuracy score. The detection accuracy score can be a score characterizing the accuracy of the scene data detection. Determining the detection accuracy score of the scene data based on the third object detection result can be achieved by calculating the accuracy of the third object detection result, thereby determining the detection accuracy score of the scene data.

[0106] In some embodiments, due to the different accuracies of the target detection model, point cloud detection model, and radar detection model, different prediction results may be generated when detecting the same scene data. If the target detection model and its corresponding point cloud detection model and radar detection model have the same judgment result for a certain obstacle in the same scene data, it indicates that the detection difficulty of the scene is low, and vice versa. Therefore, the detection difficulty score can be used to characterize the detection difficulty of the scene data. The detection difficulty score can be determined by calculating the first target detection result, the second target detection result, and the third target detection result.

[0107] In some embodiments, the weighted detection accuracy score and the weighted detection difficulty score can be summed to obtain the value score of the scene data, as shown in formula (1). Here, constants c1 and c2 are weight values, which can be the output of the target detection model, i.e., hyperparameters. They can also be customized according to actual conditions, i.e., determined in response to user operations. It should be noted that since a higher value score indicates a better training effect on the target detection model using the scene data, in order to achieve the above effect, in one or more embodiments of this application, c1 and c2 are negative numbers.

[0108] In this way, the scene data is subjected to object detection by the object detection model to obtain the third object detection result. Then, the value score of the scene data is determined based on the detection accuracy score and the detection difficulty score. The scene accuracy score is obtained based on the third object detection result, and the detection difficulty score is obtained based on the first object detection result, the second object detection result, and the third object detection result. This improves the accuracy of the value score, thereby determining the training samples of the object detection model more accurately and efficiently.

[0109] To more accurately and efficiently determine the training samples for the object detection model, as another implementation method of this application, the third object detection result includes the bounding box and the corresponding confidence score. Based on the third object detection result, the detection accuracy score of the scene data is determined, which may specifically include:

[0110] The number of bounding boxes contained in the third object detection result is determined as the number of bounding boxes.

[0111] Based on the number of bounding boxes and the confidence scores of the corresponding bounding boxes, the detection accuracy score for this scene data is determined using the following formula:

[0112]

[0113] Among them, S confidence To score the accuracy of the detection, nobj For the number of target boxes, ss camera This represents the confidence level corresponding to the target bounding box.

[0114] In the above implementation, the target box refers to the annotation of the detected obstacles by the detection model. Specifically, it can represent the location of the detected obstacles in the current scene. The confidence level of the target box is the accuracy of whether there are obstacles in the current target box. When there are indeed obstacles in the target box in the scene, it means that the confidence level of the target box is high.

[0115] The detection accuracy score of the scene data can be calculated by summing the confidence scores of all target boxes in the third detection result and then dividing the sum by the number of target boxes, as shown in formula (2).

[0116] Understandably, the lower the detection accuracy score calculated using the above formula, the more valuable the data in that scene is for training, meaning the higher the value of the scene data.

[0117] Thus, by determining the number of target boxes contained in the third target detection result as the target box count, and then determining the detection accuracy score of the scene data based on the target box count and the confidence level of the target box, since the target box confidence level is strongly correlated with the detection accuracy score, the detection accuracy score can be determined more accurately, thereby determining the training samples of the target detection model more accurately and efficiently.

[0118] To more accurately and efficiently determine the training samples for the object detection model, as another implementation method of this application, a detection difficulty score is determined based on the first object detection result, the second object detection result, and the third object detection result, specifically including:

[0119] When the input data for the object detection model is a camera image, the second object detection result is determined as the comparison object detection result.

[0120] When the input data for the target detection model is radar point cloud, the first target detection result is determined as the comparison target detection result.

[0121] Based on the detection results of the first target and the second target, the difficulty score of the first detection is determined.

[0122] The difficulty score for the second detection is determined based on the results of the third target detection and the comparison of the target detection results.

[0123] Based on the first and second detection difficulty scores, the detection difficulty score of the scene data is determined using the following formula:

[0124] S rarity =αLSrarity +βSS rarity (3)

[0125] Among them, S rarity To detect difficulty scores, α and β are constants, LS rarity The first test difficulty score is SS. rarity The second test difficulty score is given.

[0126] In the above implementation, the comparison target detection result can be a target detection result used for comparison with the third target detection result. When the comparison detection result is consistent with the third target detection result, it indicates that the detection result for this scene is relatively accurate, i.e., the detection difficulty is low, and vice versa. The first detection difficulty score can be a detection difficulty score calculated based on the first target detection result and the second target detection result. When the first target detection result is consistent with the second target detection result, it indicates that the detection result for this scene is relatively accurate, i.e., the detection difficulty is low, and vice versa. The second detection difficulty score can be a detection difficulty score calculated based on the third target detection result and the comparison target detection result corresponding to the data type of the current scene data.

[0127] In some embodiments, the first detection difficulty score and the second detection difficulty score can be weighted and summed, and the calculated value can be used as the detection difficulty score of the scene data, as shown in formula (3). In this case, α and β are constants, and their values ​​can be the output results of the target detection model, i.e., hyperparameters, or they can be customized according to the actual situation, i.e., in response to the user's operation, the values ​​of α and β are determined.

[0128] Understandably, the lower the detection difficulty score, the more valuable the data in that scene is for training, meaning the higher the value of the scene data.

[0129] Thus, the detection difficulty score of the scene data is determined by the first detection difficulty score and the second detection difficulty score. The first detection difficulty score is determined by the detection results of the first target and the second target, and the second detection difficulty score is determined by the detection results of the third target and the comparison target detection results. The detection difficulty score determined in this way includes the detection results of multiple detection models, making the score more accurate, thereby determining the training samples of the target detection model more accurately and efficiently.

[0130] To more accurately and efficiently determine the training samples for the object detection model, as another implementation of this application embodiment, a first object detection result includes a bounding box and its corresponding confidence score; a second object detection result includes a bounding box and its corresponding confidence score; and a first detection difficulty score is determined based on the first and second object detection results, specifically including:

[0131] The number of bounding boxes contained in the first object detection result is determined as the number of bounding boxes.

[0132] From the target boxes contained in the first target detection result, the target boxes that match any target box contained in the second target detection result are determined as matching target boxes.

[0133] The number of matching target boxes, the confidence level of the matching target boxes in the first target detection result, and the matching confidence level of the target boxes that match the matching target boxes in the second target detection result are determined.

[0134] The sum of the confidence scores of the matched target boxes and the matching confidence scores is used to determine the matching value.

[0135] Based on the number of matched bounding boxes, the total number of bounding boxes, and the matching value, the first detection difficulty score for this scene data is determined using the following formula:

[0136]

[0137] Among them, LS rarity The first test difficulty score is given by nll. match To match the number of bounding boxes, lsmatch i Let n be the matching value of the i-th target box. lidar This represents the number of target bounding boxes.

[0138] In the above implementation, a target box that matches any target box in the second target detection result is determined as a matching target box from the target boxes included in the first target detection result. Specifically, the target box in the second target detection result is determined to be a matching target box when the three-dimensional overlap volume of the target boxes in the first target detection result and the second target detection result is greater than or equal to a preset threshold. The preset threshold can be customized according to actual conditions, and the three-dimensional overlap volume is a parameter obtained by comparing the two target boxes in the two target detection results. Of course, since the technology for determining whether at least two target boxes match is relatively mature, the electronic device can also use other methods to determine whether the target boxes in the first target detection result and the second target detection result match. For the sake of brevity, this application will not elaborate on the specific determination methods here. Furthermore, other steps or embodiments in this application can also follow the content of the above embodiments when matching at least two target boxes.

[0139] As an example, the first object detection result contains three bounding boxes: A, B, and C. Bounding box A matches one of the bounding boxes in the second object detection result; therefore, bounding box A is the matched bounding box, and its confidence score is the confidence score of the matched bounding box A. In this case, the bounding box in the second object detection result is bounding box A2, and its confidence score is the matching confidence score.

[0140] In some embodiments, the target box and the matching target box represent the first detection result and the second detection result obtained by judging the same position in the scene. If both the first detection result and the second detection result judge that there is an obstacle in the target box and the judgment result is correct, then the confidence of the target box and the matching target box are both high. Therefore, the sum of the confidence of the matching target box and the matching confidence is determined as the matching value. The higher the matching value, the higher the detection confidence for the scene, that is, the lower the detection difficulty for the scene. The matching values ​​of all matching target boxes can be summed and then divided by the number of target boxes to obtain the first detection difficulty score, as shown in formula (4).

[0141] Thus, by calculating the number of matching target boxes, the number of target boxes, and the matching value, the first detection difficulty score is obtained. The matching value is obtained by summing the confidence of the matching target boxes and the matching confidence, which can more accurately and efficiently determine the training samples of the target detection model.

[0142] To more accurately and efficiently determine the training samples for the object detection model, as another implementation method of this application, a third object detection result, including the bounding box and the corresponding confidence score, is compared with the object detection result, including the bounding box and the corresponding confidence score. Based on the third object detection result and the compared object detection result, a second detection difficulty score is determined, specifically including:

[0143] The number of bounding boxes contained in the object detection results is determined as the number of bounding boxes.

[0144] From the bounding boxes contained in the comparison target detection results, the bounding boxes that match any bounding box contained in the third target detection results are identified as similar bounding boxes.

[0145] Determine the number of similar bounding boxes, the confidence level of similar bounding boxes in the comparative object detection results, and the similarity confidence level of bounding boxes that match similar bounding boxes in the third object detection results.

[0146] The similarity value is determined by summing the confidence scores of similar bounding boxes with the similarity scores.

[0147] Based on the number of similar bounding boxes, the total number of bounding boxes, and the similarity value, the first detection difficulty score for this scene data is determined using the following formula:

[0148]

[0149] Among them, SS rarity For the second detection difficulty score, NLS match For the number of similar bounding boxes, ssmatch i Let nl be the similarity value of the i-th similar bounding box. camera The number of target boxes.

[0150] The above implementation method is similar to the process of determining the difficulty score of the first detection, and will not be described again here.

[0151] Thus, the second detection difficulty score is calculated by the number of similar target boxes, the number of target boxes, and the similarity value. The similarity value is obtained by the sum of the confidence of similar target boxes and the similarity confidence, which can more accurately and efficiently determine the training samples of the target detection model.

[0152] To more accurately and efficiently determine the training samples for the object detection model, as another implementation of this application, the training data for the object detection model is determined based on the first n scene data in the first sequence, specifically including:

[0153] The first n scene data in the first sequence are labeled to obtain multiple training data.

[0154] The target detection model is trained using multiple training datasets.

[0155] In the above specific implementation, annotation can be performed manually on the first n scene data in the first sequence to obtain multiple training data and train the target detection model.

[0156] In this way, by labeling the unlabeled scene data, the selected scene data can be used as input data to train the object detection model, which can more accurately and efficiently determine the training samples of the object detection model.

[0157] To more accurately and efficiently determine the training samples for the object detection model, as another implementation of this application, after determining the training data for the object detection model based on the first n scene data in the first sequence, the method further includes:

[0158] Obtain the collection time identifier of the training data.

[0159] The training data is sorted according to the order of the collection time markers to obtain the second sequence.

[0160] Each pair of adjacent training data in the second sequence is determined as a training sample group.

[0161] The training data that ranks higher in the two training data sets contained in the training sample group is determined as the first training data, and the training data that ranks lower is determined as the second training data.

[0162] For each training sample group in the second sequence, when it is determined that the structural similarity between two training data in the training sample group is greater than a preset similarity threshold, the second training data in the training sample group is deleted until the structural similarity between two training data in any training sample group in the second sequence is less than or equal to the similarity threshold.

[0163] In the above implementation, the collection time identifier can be an identifier used to represent the time when the training data is collected. The collection time identifier of the training data is obtained; the training data is sorted according to the order of the collection time identifiers to obtain the second sequence, and two adjacent training data are determined as the training sample group.

[0164] In some embodiments, for each training sample group, a preset similarity threshold can represent the structural similarity between two training data. It can be set according to the actual situation or calculated using formulas. As an example, the similarity threshold is shown in formulas (6), (7), (8) and (9):

[0165]

[0166]

[0167]

[0168]

[0169] Where, for the first training data x and the second training data y, μ x Let μ be the mean of x. y Let σ be the mean of y. x 2 Let σ be the variance of x. y 2 Let σ be the variance of y. xy Let C1 be the covariance of x and y, and C1 = (k1L) 2 C2=(k2L) 2 C3 = C2 / 2 are three constants used to avoid division by zero. L is the dynamic range of the image pixel values. k1 and k2 are constants much less than 1. l(x,y) is the brightness contrast function, c(x,y) is the contrast contrast function, s(x,y) is the structure contrast function, and S(x,y) is the contrast function.

[0170] It should be noted that the structural similarity between two training data sets can be calculated using algorithms or models such as the Self-Similarity Matrix (SSM) algorithm and the Perceptual Hash algorithm. This application does not limit the specific algorithm or model used and it can be set as needed. The similarity threshold can be set in response to user operations or determined based on the average, median, or other values ​​of the structural similarity of the training data. This application does not restrict how the similarity threshold is specifically determined and it can be set as needed.

[0171] In this way, by calculating the similarity of the data for each training sample and filtering out the data with high similarity, highly repetitive scenes can be removed, thereby determining the training samples for the target detection model more accurately and efficiently.

[0172] like Figure 2 As shown, Figure 2 This is a flowchart illustrating another data processing method provided in one embodiment of this application. Specifically, it includes the following steps:

[0173] S201, collect scene data.

[0174] S202, Calculate the detection result of the first target.

[0175] S203, Calculate the detection result of the second target.

[0176] S204, Calculate the detection result of the third target.

[0177] S205, Calculated Value Score.

[0178] S206, Filter data.

[0179] S207 Select training sample data.

[0180] Furthermore, the acquisition, storage, use, and processing of data in this application's technical solution all comply with relevant national laws and regulations.

[0181] Based on the data processing method provided in the above embodiments, this application also provides specific implementations of a data processing apparatus. Please refer to the following embodiments.

[0182] First see Figure 3 The data processing apparatus provided in this application includes the following units:

[0183] The acquisition unit 301 is used to acquire multiple scene data collected by the vehicle during driving. The scene data is collected by the camera sensor and radar sensor configured in the vehicle.

[0184] The detection unit 302 is used to perform target detection on each scene data using a point cloud detection model to obtain a first target detection result; and to perform target detection on the scene data using an image detection model to obtain a second target detection result.

[0185] The scoring unit 303 is used to determine the value score of the scene data based on the first target detection result and the second target detection result;

[0186] Sorting unit 304 is used to sort multiple scenario data in descending order of value score to obtain the first sequence;

[0187] The determination unit 305 is used to determine the training data of the target detection model based on the first n scene data in the first sequence.

[0188] The above describes the specific implementation of the data processing method provided in this application embodiment. It can be seen that in the above embodiment, the determining unit 305 can determine the training data of the target detection model based on several scene data with high value scores. The value score is determined by the scoring unit 303 based on the first target detection result and the second target detection result. The first target detection result is obtained by the detection unit 302 through target detection of the scene data using a point cloud detection model, and the second target detection result is obtained by the detection unit 302 through target detection of the scene data using an image detection model. Therefore, by using scene data collected from multiple sensors, scene data with high value scores can be determined more accurately and efficiently, thereby determining training samples more accurately and efficiently.

[0189] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include a scoring subunit 3031.

[0190] The scoring subunit 3031 is used to perform target detection on the scene data using a target detection model to obtain a third target detection result. The input data for the target detection model is radar point cloud or camera image. Based on the third target detection result, the detection accuracy score of the scene data is determined. Based on the first, second, and third target detection results, the detection difficulty score is determined. Based on the detection accuracy score and the detection difficulty score, the value score of the scene data is determined using the following formula:

[0191] S = c1S confidence +c2S rarity

[0192] Where S is the value rating, and c1 and c2 are constants, S confidenceTo score the accuracy of the detection, S rarity To assess the difficulty level of the test.

[0193] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include a scoring subunit 3032.

[0194] The scoring subunit 3032 is used to determine the number of bounding boxes contained in the third target detection result as the number of bounding boxes. Based on the number of bounding boxes and the confidence level of each bounding box, the detection accuracy score of the scene data is determined using the following formula:

[0195]

[0196] Among them, S confidence To score the accuracy of the detection, n obj For the number of target boxes, ss camera This represents the confidence level corresponding to the target bounding box.

[0197] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include a scoring subunit 3033.

[0198] The scoring subunit 3033 is used to determine the second target detection result as the comparison target detection result when the input data of the target detection model is a camera image, and to determine the first target detection result as the comparison target detection result when the input data of the target detection model is a radar point cloud. Based on the first and second target detection results, a first detection difficulty score is determined; based on the third and comparison target detection results, a second detection difficulty score is determined; and based on the first and second detection difficulty scores, the scene data detection difficulty score is determined using the following formula:

[0199] S rarity =αLS rarity +βSS rarity

[0200] Among them, S rarity To detect difficulty scores, α and β are constants, LS rarity The first test difficulty score is SS. rarity The second test difficulty score is given.

[0201] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include a scoring subunit 3034.

[0202] The scoring subunit 3034 is used to determine the number of target boxes contained in the first target detection result as the number of target boxes; from the target boxes contained in the first target detection result, the target boxes that match any target box contained in the second target detection result are determined as matching target boxes; the number of matching target boxes, the confidence level of the matching target boxes in the first target detection result, and the matching confidence level of the target boxes that match the matching target boxes in the second target detection result are determined; the sum of the confidence level of the matching target boxes and the matching confidence level is determined as the matching value; and based on the number of matching target boxes, the number of target boxes, and the matching value, the first detection difficulty score of the scene data is determined by the following formula:

[0203]

[0204] Among them, LS rarity The first test difficulty score is given by nll. match To match the number of bounding boxes, lsmatch i Let n be the matching value of the i-th target box. lidar This represents the number of target bounding boxes.

[0205] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include a scoring subunit 3035.

[0206] The scoring subunit 3035 is used to determine the number of target boxes contained in the comparison target detection results as the target box count; from the target boxes contained in the comparison target detection results, the target boxes that match any target box contained in the third target detection results are determined as similar target boxes; the number of similar target boxes, the confidence level of the similar target boxes in the comparison target detection results, and the similarity confidence level of the target boxes that match the similar target boxes in the third target detection results are determined; the sum of the confidence levels of the similar target boxes and the similarity confidence levels is determined as the similarity value; based on the number of similar target boxes, the target box count, and the similarity value, the first detection difficulty score of the scene data is determined using the following formula:

[0207]

[0208] Among them, SS rarity For the second detection difficulty score, NLS match For the number of similar bounding boxes, ssmatch i Let nl be the similarity value of the i-th similar bounding box. camera The number of target boxes.

[0209] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include: a determination subunit 3051.

[0210] Subunit 3051 is defined to label the first n scene data in the first sequence to obtain multiple training data. The target detection model is trained using the multiple training data.

[0211] As another implementation of this application, in order to determine the training samples of the target detection model more accurately and efficiently, the above-mentioned device may further include: a determination subunit 3052.

[0212] Subunit 3052 is used to obtain the collection time identifier of the training data. The training data is sorted according to the order of the collection time identifier to obtain a second sequence. Each pair of adjacent training data in the second sequence is determined as a training sample group. The training data that is ranked earlier in the training sample group is determined as the first training data, and the training data that is ranked later is determined as the second training data. For each training sample group in the second sequence, when it is determined that the structural similarity between the two training data in the training sample group is greater than a preset similarity threshold, the second training data in the training sample group is deleted, until the structural similarity between the two training data in any training sample group in the second sequence is less than or equal to the similarity threshold.

[0213] Figure 4 A schematic diagram of the hardware structure for data processing provided in an embodiment of this application is shown.

[0214] The data processing device may include a processor 401 and a memory 402 storing computer program instructions.

[0215] Specifically, the processor 401 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0216] Memory 402 may include mass storage for data or instructions. For example, and not limitingly, memory 402 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 402 may include removable or non-removable (or fixed) media. Where appropriate, memory 402 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 402 is non-volatile solid-state memory.

[0217] In a particular embodiment, memory 402 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Thus, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure.

[0218] The processor 401 implements any of the data processing methods described in the above embodiments by reading and executing computer program instructions stored in the memory 402.

[0219] In one example, the data processing device may further include a communication interface 403 and a bus 410. Wherein, as... Figure 4 As shown, the processor 401, memory 402, and communication interface 403 are connected through bus 410 and complete communication with each other.

[0220] The communication interface 403 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0221] Bus 410 includes hardware, software, or both, that couples components of a data processing device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 410 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0222] The data processing device can execute the data processing method described in this application embodiment based on currently blocked spam SMS messages and SMS messages reported by users, thereby achieving a combination of... Figure 1 and Figure 3 The data processing methods and apparatus described.

[0223] Furthermore, in conjunction with the data processing methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the data processing methods in the above embodiments.

[0224] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0225] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0226] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0227] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0228] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A data processing method, characterized in that, include: The system acquires multiple scene data collected during vehicle operation, the scene data being collected by camera sensors and radar sensors configured in the vehicle. For each scene data, a point cloud detection model is used to perform target detection on the scene data to obtain the first target detection result; The scene data is used to perform target detection using an image detection model to obtain a second target detection result; Based on the first target detection result and the second target detection result, determine the value score of the scene data; The data from multiple scenarios are sorted in descending order of value scores to obtain the first sequence. Based on the first n scene data in the first sequence, determine the training data for the target detection model; The step of determining the value score of the scene data based on the first target detection result and the second target detection result specifically includes: The scene data is subjected to target detection by a target detection model to obtain a third target detection result. The input data of the target detection model is radar point cloud or camera image. Based on the detection results of the third target, a detection accuracy score for the scene data is determined; Based on the detection results of the first target, the second target, and the third target, a detection difficulty score is determined. Based on the detection accuracy score and the detection difficulty score, the value score of the scene data is determined using the following formula: Where S is the value score. as well as It is a constant. The accuracy of the detection is scored. The difficulty level of the detection is scored; The third target detection result includes the target bounding box and the confidence score corresponding to the target bounding box. The step of determining the detection accuracy score of the scene data based on the third target detection result specifically includes: The number of target boxes contained in the third target detection result is determined as the number of target boxes; Based on the number of bounding boxes and the confidence level of each bounding box, the detection accuracy score for this scene data is determined using the following formula: in, The accuracy of the detection is scored. The number of target boxes. The confidence level is the value corresponding to the target bounding box.

2. The method according to claim 1, characterized in that, The step of determining the detection difficulty score based on the first target detection result, the second target detection result, and the third target detection result specifically includes: When the input data of the target detection model is a camera image, the second target detection result is determined as the comparison target detection result; When the input data of the target detection model is radar point cloud, the first target detection result is determined as the comparison target detection result; Based on the first target detection result and the second target detection result, a first detection difficulty score is determined; Based on the third target detection result and the comparison target detection result, a second detection difficulty score is determined; Based on the first detection difficulty score and the second detection difficulty score, the detection difficulty score of the scene data is determined using the following formula: in, The difficulty level of the detection is scored. as well as It is a constant. The first test difficulty score is given. The second test difficulty score is given.

3. The method according to claim 2, characterized in that, The first target detection result includes a bounding box and a corresponding confidence level; the second target detection result includes a bounding box and a corresponding confidence level; and determining a first detection difficulty score based on the first and second target detection results specifically includes: The number of target boxes contained in the first target detection result is determined as the number of target boxes; From the target boxes included in the first target detection result, the target boxes that match any target box included in the second target detection result are determined as matching target boxes; The number of matching target boxes, the confidence level of the matching target boxes in the first target detection result, and the matching confidence level of the target boxes that match the matching target boxes in the second target detection result are determined. The matching value is determined by summing the confidence score of the matching target box with the matching confidence score. Based on the number of matching bounding boxes, the number of bounding boxes, and the matching value, the first detection difficulty score for this scene data is determined using the following formula: in, The difficulty level of the first detection is scored. The number of the matched target boxes. Let be the matching value of the i-th matching target box. The number of target boxes.

4. The method according to claim 2, characterized in that, The third target detection result includes a bounding box and the confidence level corresponding to the bounding box; the comparison target detection result includes a bounding box and the confidence level corresponding to the bounding box; and the step of determining a second detection difficulty score based on the third target detection result and the comparison target detection result specifically includes: The number of target boxes contained in the comparison target detection results is determined as the number of target boxes; From the target boxes included in the comparison target detection results, the target boxes that match any target box included in the third target detection results are determined as similar target boxes; The number of similar target boxes, the confidence level of the similar target boxes in the comparison target detection results, and the similarity confidence level of the target boxes that match the similar target boxes in the third target detection results are determined. The similarity value is determined by summing the confidence scores of the similar bounding boxes and the similarity scores. Based on the number of similar bounding boxes, the total number of bounding boxes, and the similarity value, the first detection difficulty score for this scene data is determined using the following formula: in, The second detection difficulty score is given. The number of such target boxes. Let be the similarity value for the i-th similar bounding box. The number of target boxes.

5. The method according to claim 1, characterized in that, The step of determining the training data for the target detection model based on the first n scene data in the first sequence specifically includes: The first n scene data in the first sequence are labeled to obtain multiple training data; The target detection model is trained using the aforementioned training data.

6. The method according to claim 1, characterized in that, After determining the training data for the object detection model based on the first n scene data in the first sequence, the method further includes: Obtain the collection time identifier of the training data; The training data is sorted according to the order of the collection time identifiers to obtain a second sequence; Each pair of adjacent training data in the second sequence is determined as a training sample group; The training data that ranks first among the two training data contained in the training sample group is determined as the first training data, and the training data that ranks last is determined as the second training data. For each training sample group in the second sequence, when it is determined that the structural similarity between two training data in the training sample group is greater than a preset similarity threshold, the second training data in the training sample group is deleted until the structural similarity between two training data in any training sample group in the second sequence is less than or equal to the similarity threshold.

7. A data processing apparatus, characterized in that, The device includes: The acquisition unit is used to acquire multiple scene data collected by the vehicle during driving, the scene data being collected by the camera sensor and radar sensor configured in the vehicle; The detection unit is used to perform target detection on each scene data using a point cloud detection model to obtain a first target detection result; and to perform target detection on the scene data using an image detection model to obtain a second target detection result. The scoring unit is used to determine the value score of the scene data based on the first target detection result and the second target detection result; The sorting unit is used to sort multiple scenario data in descending order of value score to obtain the first sequence; The determining unit is used to determine the training data of the target detection model based on the first n scene data in the first sequence; The scoring unit specifically includes: The scene detection subunit is used to perform target detection on the scene data through the target detection model to obtain a third target detection result. The input data of the target detection model is radar point cloud or camera image. The accuracy scoring subunit is used to determine the detection accuracy score of the scene data based on the detection result of the third target. The difficulty scoring subunit is used to determine the detection difficulty score based on the first target detection result, the second target detection result, and the third target detection result; The value scoring subunit is used to determine the value score of the scene data based on the detection accuracy score and the detection difficulty score, using the following formula: Where S is the value score. as well as It is a constant. The accuracy of the detection is scored. The difficulty level of the detection is scored; The third target detection result includes the target bounding box and the confidence level corresponding to the target bounding box. The accuracy scoring subunit specifically includes: The target bounding box sub-unit is used to determine the number of target bounding boxes contained in the third target detection result as the number of target bounding boxes; The bounding box calculation subunit is used to determine the detection accuracy score of the scene data based on the number of bounding boxes and the confidence level of each bounding box, using the following formula: in, The accuracy of the detection is scored. The number of target boxes. The confidence level is the value corresponding to the target bounding box.

8. A data processing device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the data processing method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the data processing method as described in any one of claims 1-6.

10. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the data processing method as described in any one of claims 1-6.

Images