An identity recognition method and device, an electronic device, and a storage medium

Abstract

The application discloses an identity recognition method and device, electronic equipment and storage medium, and the method comprises the steps of determining to-be-recognized gait feature information and to-be-recognized attribute information of a target object in a to-be-processed video; matching the to-be-recognized gait feature information with candidate gait feature information of a candidate object, determining a first similarity between the target object and the candidate object; matching the to-be-recognized attribute information with candidate attribute information of the candidate object, correcting the first similarity according to a matching result, determining a second similarity between the target object and the candidate object; and determining whether the target object is the candidate object according to the second similarity. The embodiment of the application corrects the first similarity by using the to-be-recognized attribute information, determines whether the target object is the candidate object according to the second similarity obtained after correction, and improves the accuracy of identity recognition of the target object.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biometric identification, and particularly relates to an identity recognition method and device, an electronic device and a storage medium. BACKGROUND

[0002] Gait recognition is a new biometric identification technology, which identifies the identity through the posture of people walking, and has many advantages such as non-contact and long-distance recognition. Especially in today's increasingly developed video monitoring technology, gait recognition technology is becoming more and more the focus of researchers. The current mainstream gait recognition algorithm mostly uses a deep learning model to obtain a pedestrian gait contour sequence, and then extracts features from the pedestrian gait contour sequence to obtain to-be-identified gait feature information, matches the to-be-identified gait feature information with candidate gait feature information in a feature library, so as to realize the identity recognition of the pedestrian. However, the misjudgment rate of identity recognition by relying only on gait feature information is high, and the accuracy of identity recognition is poor. SUMMARY

[0003] Embodiments of the present application provide an identity recognition method, device, electronic device and storage medium, to solve the problem of poor accuracy of existing identity recognition.

[0004] An identity recognition method is provided in an embodiment of the present application, and the method comprises:

[0005] determining to-be-identified gait feature information and to-be-identified attribute information of a target object in a to-be-processed video;

[0006] matching the to-be-identified gait feature information with candidate gait feature information of a candidate object, to determine a first similarity between the target object and the candidate object;

[0007] matching the to-be-identified attribute information with candidate attribute information of the candidate object, correcting the first similarity according to a matching result, to determine a second similarity between the target object and the candidate object;

[0008] determining whether the target object is the candidate object according to the second similarity.

[0009] Further, the determination of the to-be-identified gait feature information of the target object in the to-be-processed video comprises:

[0010] inputting each image in the to-be-processed video into a trained image segmentation model respectively, determining to-be-identified gait contour images of the target object in each image based on the image segmentation model, and determining a to-be-identified gait contour sequence according to each to-be-identified gait contour image; wherein the image segmentation model is obtained by training a first sample image labeled with gait contour position information.

[0011] inputting the to-be-identified gait profile sequence into the trained gait feature extraction model, and determining to-be-identified gait feature information corresponding to the to-be-identified gait profile sequence based on the gait feature extraction model, wherein the gait feature extraction model is obtained by training sample gait profile sequences labeled with sample gait feature information.

[0012] Further, determining the to-be-identified attribute information of the target object in the to-be-processed video comprises:

[0013] determining a to-be-identified image in the to-be-processed video, inputting the to-be-identified image into the trained attribute extraction model, and determining the to-be-identified attribute information of the target object in the to-be-identified image based on the attribute extraction model, wherein the attribute extraction model is obtained by training the second sample image labeled with sample object attribute information.

[0014] Further, the determining whether the target object is the candidate object according to the second similarity comprises:

[0015] If it is determined that the second similarity is greater than a preset first similarity threshold, it is determined that the target object is the candidate object.

[0016] Further, the candidate object comprises at least two, and the determining whether the target object is the candidate object according to the second similarity comprises:

[0017] determining a second similarity between the target object and each candidate object, selecting a candidate object corresponding to a maximum value of the determined second similarities, and determining that the target object is the selected candidate object.

[0018] Further, the selecting a candidate object corresponding to a maximum value of the determined second similarities and determining that the target object is the selected candidate object comprises:

[0019] selecting a candidate object corresponding to a maximum second similarity from second similarities greater than a preset second similarity threshold, and determining that the target object is the selected candidate object.

[0020] Further, the attribute information comprises age information, gender information, and skin color information.

[0021] The matching the to-be-identified attribute information with candidate attribute information of the candidate object, correcting the first similarity according to a matching result, and determining a second similarity between the target object and the candidate object comprises:

[0022] For each to-be-identified attribute information, the to-be-identified attribute information is matched with corresponding candidate attribute information of the candidate object in the feature library to determine a similarity change value corresponding to the to-be-identified attribute information;

[0023] According to the similarity change value corresponding to each to-be-identified attribute information and a predetermined weight value corresponding to each to-be-identified attribute information, a similarity change value of the target object and the candidate object is determined.

[0024] According to the similarity change value, the first similarity is corrected to determine a second similarity of the target object and the candidate object.

[0025] Further, after the first similarity of the target object and the candidate object is determined, before the to-be-identified attribute information is matched with the candidate attribute information of the candidate object, the method further comprises:

[0026] If the first similarity is greater than a predetermined third similarity threshold, it is determined that the target object is the candidate object; if the first similarity is less than a predetermined fourth similarity threshold, it is determined that the target object is not the candidate object.

[0027] If the first similarity is between the predetermined third similarity threshold and the predetermined fourth similarity threshold, the step of matching the to-be-identified attribute information with the candidate attribute information of the candidate object is performed; wherein the predetermined third similarity threshold is greater than the predetermined fourth similarity threshold.

[0028] On the other hand, an embodiment of the present application provides an identity recognition device, the device comprising:

[0029] A first determination module is configured to determine to-be-identified gait feature information and to-be-identified attribute information of a target object in a to-be-processed video.

[0030] A second determination module is configured to match the to-be-identified gait feature information with candidate gait feature information of a candidate object to determine a first similarity of the target object and the candidate object.

[0031] A third determination module is configured to match the to-be-identified attribute information with candidate attribute information of the candidate object, correct the first similarity according to a matching result, and determine a second similarity of the target object and the candidate object.

[0032] A recognition module is configured to determine whether the target object is the candidate object according to the second similarity.

[0033] Further, the first determining module is specifically configured to input each image in the to-be-processed video into a trained image segmentation model respectively, determine a to-be-identified gait contour graph of the target object in each image based on the image segmentation model, and determine a to-be-identified gait contour sequence according to each to-be-identified gait contour graph; wherein the image segmentation model is obtained by training first sample images labeled with gait contour position information; input the to-be-identified gait contour sequence into a trained gait feature extraction model, and determine to-be-identified gait feature information corresponding to the to-be-identified gait contour sequence based on the gait feature extraction model; wherein the gait feature extraction model is obtained by training sample gait contour sequences labeled with sample gait feature information.

[0034] Further, the first determining module is specifically configured to determine a to-be-identified image in the to-be-processed video, input the to-be-identified image into a trained attribute extraction model, and determine to-be-identified attribute information of the target object in the to-be-identified image based on the attribute extraction model; wherein the attribute extraction model is obtained by training second sample images labeled with sample object attribute information.

[0035] Further, the identifying module is specifically configured to determine that the target object is the candidate object if it is determined that the second similarity is greater than a preset first similarity threshold.

[0036] Further, the identifying module is specifically configured to determine a second similarity between the target object and each candidate object, select a candidate object corresponding to a maximum value of the determined second similarities, and determine that the target object is the selected candidate object.

[0037] Further, the identifying module is specifically configured to select a candidate object corresponding to a maximum second similarity from second similarities greater than a preset second similarity threshold, and determine that the target object is the selected candidate object.

[0038] Further, the third determining module is specifically configured to, for each to-be-identified attribute information, match the to-be-identified attribute information with corresponding candidate attribute information of the candidate object in a feature library, determine a similarity change value corresponding to the to-be-identified attribute information, determine a similarity change value between the target object and the candidate object according to the similarity change value corresponding to each to-be-identified attribute information and a weight value corresponding to each to-be-identified attribute information determined in advance, and correct the first similarity according to the similarity change value to determine a second similarity between the target object and the candidate object.

[0039] Further, the identification module is further configured to: if the first similarity is greater than a third preset similarity threshold, determine that the target object is the candidate object; if the first similarity is less than a fourth preset similarity threshold, determine that the target object is not the candidate object; and if the first similarity is between the third preset similarity threshold and the fourth preset similarity threshold, trigger the third determination module, wherein the third preset similarity threshold is greater than the fourth preset similarity threshold.

[0040] In another aspect, an electronic device is provided, which includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus.

[0041] The memory is configured to store a computer program.

[0042] The processor is configured to execute the program stored in the memory, and implement the method steps of any of the above aspects.

[0043] In another aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the method steps of any of the above aspects.

[0044] The embodiments of the present application provide an identity recognition method and device, an electronic device and a storage medium, and the method includes: determining to-be-recognized gait feature information and to-be-recognized attribute information of a target object in a to-be-processed video; matching the to-be-recognized gait feature information with candidate gait feature information of a candidate object, to determine a first similarity between the target object and the candidate object; matching the to-be-recognized attribute information with candidate attribute information of the candidate object, and correcting the first similarity according to a matching result, to determine a second similarity between the target object and the candidate object; and determining whether the target object is the candidate object according to the second similarity.

[0045] The above technical solution has the following advantages or beneficial effects:

[0046] In the embodiments of the present application, the to-be-recognized gait feature information and the to-be-recognized attribute information of the target object in the to-be-processed video are determined, the first similarity between the target object and the candidate object is determined based on the to-be-recognized gait feature information, the first similarity between the target object and the candidate object is corrected based on the to-be-recognized attribute information, to obtain the second similarity, and then the identity recognition is performed according to the second similarity. In the embodiments of the present application, the to-be-recognized attribute information is used to correct the first similarity between the target object and the candidate object determined based on the to-be-recognized gait feature information, and the second similarity obtained after the correction is used to determine whether the target object is the candidate object, thereby improving the accuracy of the identity recognition of the target object. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of the identity recognition process provided in Embodiment 1 of the present invention;

[0049] Figure 2 This is a schematic diagram of identity recognition provided in Embodiment 7 of the present invention;

[0050] Figure 3 This is a schematic diagram of the identity recognition device provided in Embodiment 8 of the present invention;

[0051] Figure 4 This is a schematic diagram of the electronic device structure provided in Embodiment 9 of the present invention. Detailed Implementation

[0052] The present invention will now be described in further detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0053] Example 1:

[0054] Figure 1 This is a schematic diagram of an identity recognition process provided in an embodiment of the present invention. The process includes the following steps:

[0055] S101: Determine the gait features and attribute information to be identified for the target object in the video to be processed.

[0056] S102: Match the gait feature information to be identified with the candidate gait feature information of the candidate object to determine the first similarity between the target object and the candidate object.

[0057] S103: Match the attribute information to be identified with the candidate attribute information of the candidate object, correct the first similarity based on the matching result, and determine the second similarity between the target object and the candidate object.

[0058] S104: Determine whether the target object is the candidate object based on the second similarity.

[0059] The identity recognition method provided by the embodiment of the present application is applied to an electronic device, which can be a PC, a tablet computer or the like, or a camera or the like. If the electronic device is a camera, after the camera collects a to-be-processed video, the to-be-recognized gait feature information and the to-be-recognized attribute information of a target object in the to-be-processed video are determined, and a subsequent identity recognition process is performed. If the electronic device is a PC, a tablet computer or the like, after the camera collects a to-be-processed video, the to-be-processed video is first sent to the electronic device, and then the to-be-recognized gait feature information and the to-be-recognized attribute information of a target object in the to-be-processed video are determined by the electronic device, and a subsequent identity recognition process is performed.

[0060] The electronic device determines the to-be-recognized gait feature information of the target object in the to-be-processed video. The to-be-recognized gait feature information of the target object in the to-be-processed video can be determined based on an image selected from the to-be-processed video. Alternatively, the to-be-recognized gait feature information of the target object in the to-be-processed video can be determined based on a to-be-recognized gait contour sequence obtained by sorting image frames in the to-be-processed video. The to-be-recognized gait feature information of the target object in the to-be-processed video can be determined by deep learning.

[0061] The electronic device determines the to-be-recognized attribute information of the target object in the to-be-processed video. The to-be-recognized attribute information can include one or more of age information, gender information and skin color information. The to-be-recognized attribute information of the target object in the to-be-processed video can be determined by deep learning.

[0062] The feature library pre-stored in the electronic device includes candidate gait feature information and candidate attribute information of candidate objects. The to-be-recognized gait feature information is matched with the candidate gait feature information of the candidate objects in the feature library, and the to-be-recognized attribute information is matched with the candidate attribute information of the candidate objects in the feature library, so as to determine the similarity between the target object and the candidate objects, and then determine whether the target object is a candidate object according to the similarity.

[0063] Specifically, the to-be-recognized gait feature information is matched with the candidate gait feature information of the candidate objects, so as to determine the first similarity between the target object and the candidate objects. The to-be-recognized gait feature information of the target object and the candidate gait feature information of the candidate objects in the feature library can be represented as feature vectors. The to-be-recognized gait feature information is matched with the candidate gait feature information of the candidate objects in the feature library, so as to determine the cosine distance between the to-be-recognized gait feature information and the candidate gait feature information of the candidate objects in the feature library. The smaller the cosine distance is, the greater the first similarity is.

[0064] The to-be-recognized attribute information is matched with the candidate attribute information of the candidate object, the first similarity is corrected according to a matching result, and the second similarity of the target object and the candidate object is determined. The to-be-recognized attribute information is matched with the candidate attribute information of the candidate object in the feature library, that is, whether the to-be-recognized attribute information is the same as the candidate attribute information of the candidate object in the feature library is determined. The first similarity is corrected, for example, if the to-be-recognized attribute information is the same as the candidate attribute information, the first similarity is taken as the corrected second similarity, or a certain value is added to the first similarity to obtain the second similarity. If the to-be-recognized attribute information is different from the candidate attribute information, a certain value is subtracted from the first similarity to obtain the second similarity. Finally, whether the target object is the candidate object is determined according to the second similarity. It should be noted that the feature library also includes identity information of the candidate object, for example, the name and the identity card number of the candidate object. If it is determined that the target object is the candidate object, the identity information of the candidate object is determined as the identity information of the target object, and identity recognition of the target object is realized.

[0065] In the embodiment of the application, the to-be-recognized attribute information and the to-be-recognized gait feature information of the target object in the to-be-processed video are determined, the first similarity of the target object and the candidate object is determined based on the to-be-recognized gait feature information, the first similarity of the target object and the candidate object is corrected based on the to-be-recognized attribute information to obtain the second similarity, and then identity recognition is performed according to the second similarity. In the embodiment of the application, the to-be-recognized attribute information is used to correct the first similarity of the target object and the candidate object determined based on the to-be-recognized gait feature information, whether the target object is the candidate object is determined according to the second similarity obtained after correction, and the accuracy of identity recognition of the target object is improved.

[0066] Embodiment 2

[0067] In order to determine the to-be-recognized gait feature information of the target object in the to-be-processed video, on the basis of the above-mentioned embodiments, in the embodiment of the application, the determination of the to-be-recognized gait feature information of the target object in the to-be-processed video comprises:

[0068] Each image in the to-be-processed video is input into the trained image segmentation model, the to-be-recognized gait contour of the target object in each image is determined based on the image segmentation model, and a to-be-recognized gait contour sequence is determined according to each to-be-recognized gait contour. The image segmentation model is obtained by training a first sample image labeled with gait contour position information.

[0069] input the gait contour sequence to be recognized into the trained gait feature extraction model, and determine the gait feature information to be recognized corresponding to the gait contour sequence to be recognized based on the gait feature extraction model; wherein the gait feature extraction model is obtained by training sample gait contour sequences labeled with sample gait feature information.

[0070] In the embodiments of the present application, the electronic device determines the gait feature information to be recognized of the target object in the video to be processed based on the image segmentation model and the gait feature extraction model. The image segmentation model is used to determine the gait contour map to be recognized of the target object in each image in the video to be processed. The gait feature extraction model is used to determine the gait feature information to be recognized of the target object.

[0071] In training the image segmentation model, for each first sample image in the first training set, the first sample image and the corresponding labeled image are input into the image segmentation model, and the image segmentation model is trained. The labeled image is labeled with the gait contour position information of the first sample object in the corresponding first sample image. After the image segmentation model is trained, each image in the video to be processed is input into the trained image segmentation model, and the gait contour map to be recognized of the target object in each image is determined based on the image segmentation model. According to each gait contour map to be recognized, the gait contour sequence to be recognized is obtained by sorting the image frames. It should be noted that in the embodiments of the present application, in the process of determining the gait contour sequence to be recognized, the images in the video to be processed can be first screened, for example, a certain number of images with good quality are selected from the video to be processed, then the selected images are input into the trained image segmentation model, the gait contour map to be recognized of the target object in each image is determined based on the image segmentation model, and then the gait contour sequence to be recognized is determined according to each gait contour map to be recognized. Thus, the accuracy of the gait contour sequence to be recognized obtained is ensured.

[0072] In training the gait feature extraction model, for each sample gait contour sequence in the second training set, the sample gait contour sequence and the corresponding sample gait feature information are input into the gait feature extraction model, and the gait feature extraction model is trained. After the gait feature extraction model is trained, the gait contour sequence to be recognized is input into the trained gait feature extraction model, and the gait feature information to be recognized corresponding to the gait contour sequence to be recognized is determined based on the gait feature extraction model.

[0073] Embodiment 3:

[0074] In order to determine the attribute information to be recognized of the target object in the video to be processed, on the basis of the above-mentioned embodiments, in the embodiments of the present application, the determination of the attribute information to be recognized of the target object in the video to be processed comprises:

[0075] Determine a to-be-recognized image in the to-be-processed video, input the to-be-recognized image into the trained attribute extraction model, and determine to-be-recognized attribute information of the target object in the to-be-recognized image based on the attribute extraction model; wherein the attribute extraction model is obtained by training a second sample image labeled with sample object attribute information.

[0076] In the embodiment of the application, the to-be-recognized attribute information of the target object in the to-be-processed video is determined based on the trained attribute extraction model. When training the attribute extraction model, for each second sample image in the third training set, the second sample image and the corresponding second sample object attribute information are input into the attribute extraction model, and the attribute extraction model is trained. After the attribute extraction model is trained, a to-be-recognized image is determined in the to-be-processed video, the to-be-recognized image is input into the trained attribute extraction model, and the to-be-recognized attribute information of the target object in the to-be-recognized image is determined based on the attribute extraction model. The to-be-recognized attribute information of the target object in the to-be-recognized image determined is the to-be-recognized attribute information of the target object in the to-be-processed video. It should be noted that when the to-be-recognized image is determined in the to-be-processed video, the best quality image in the to-be-processed video can be used as the to-be-recognized image.

[0077] Embodiment 4:

[0078] In order to perform identity recognition, on the basis of each of the above embodiments, in the embodiment of the application, the determining whether the target object is the candidate object according to the second similarity includes:

[0079] If it is determined that the second similarity is greater than the preset first similarity threshold, it is determined that the target object is the candidate object.

[0080] The electronic device stores a preset first similarity threshold, which is, for example, 0.8, 0.85, etc. After the electronic device determines the second similarity between the target object and the candidate object, it is determined whether the second similarity is greater than the preset first similarity threshold. If yes, it is determined that the target object is the candidate object. If no, it is determined that the target object is not the candidate object.

[0081] Embodiment 5:

[0082] The candidate object in the feature library can be multiple. When the candidate object includes at least two, in order to perform identity recognition, on the basis of each of the above embodiments, in the embodiment of the application, the candidate object includes at least two, and the determining whether the target object is the candidate object according to the second similarity includes:

[0083] determining a second similarity between the target object and each candidate object; selecting a candidate object corresponding to a maximum value of the determined second similarities; and determining the target object as the selected candidate object.

[0084] In the embodiment of the present application, the candidate objects include at least two, and the to-be-recognized gait feature information and the to-be-recognized attribute information are matched with the candidate gait feature information and the candidate attribute information of each candidate object in the feature library respectively to determine a second similarity between the target object and each candidate object. For example, the feature library includes the candidate gait feature information and the candidate attribute information of a candidate object A, the candidate gait feature information and the candidate attribute information of a candidate object B, and the candidate gait feature information and the candidate attribute information of a candidate object C. The to-be-recognized gait feature information and the to-be-recognized attribute information are matched with the candidate gait feature information and the candidate attribute information of the candidate object A respectively to determine a second similarity between the target object and the candidate object A. The to-be-recognized gait feature information and the to-be-recognized attribute information are matched with the candidate gait feature information and the candidate attribute information of the candidate object B respectively to determine a second similarity between the target object and the candidate object B. The to-be-recognized gait feature information and the to-be-recognized attribute information are matched with the candidate gait feature information and the candidate attribute information of the candidate object C respectively to determine a second similarity between the target object and the candidate object C. Then, a candidate object corresponding to a maximum second similarity is selected from the three determined second similarities, and the target object is determined as the selected candidate object. For example, the second similarity between the target object and the candidate object C is the maximum, and the target object is determined as the candidate object C.

[0085] In order to further make the identity recognition more accurate, in the embodiment of the present application, the selecting a candidate object corresponding to a maximum value of the determined second similarities; and determining the target object as the selected candidate object includes:

[0086] The candidate object corresponding to the maximum second similarity is selected from the second similarities greater than the preset second similarity threshold, and the target object is determined as the selected candidate object.

[0087] The preset second similarity threshold is stored in the electronic device, and the preset second similarity threshold and the preset first similarity threshold can be the same or different. After the electronic device determines the second similarity between the target object and each candidate object, the second similarity greater than the preset second similarity threshold is selected first. Then, the maximum second similarity is selected from the selected second similarities greater than the preset second similarity threshold, and the target object is determined as the candidate object corresponding to the selected maximum second similarity.

[0088] Since in the embodiment of the present application, when the candidate objects include at least two, the second similarity between the target object and each candidate object is first determined, the second similarity greater than the preset second similarity threshold is screened out, and the maximum second similarity is selected from the screened second similarity, and the target object is determined as the candidate object corresponding to the selected maximum second similarity. Therefore, the identity recognition in the embodiment of the present application is more accurate.

[0089] It should be noted that if the second similarity between the target object and each candidate object determined does not exist the second similarity greater than the preset second similarity threshold, it is considered that there is no candidate object in the feature library identical to the target object.

[0090] In addition, one or more candidate objects with higher similarity to the target object can be selected based on the scheme provided in the embodiment of the present application according to needs. The specific selection strategy can be any one of the following: Strategy one, selecting the candidate object corresponding to the maximum second similarity from the determined second similarity between the selected target object and each candidate object. Strategy two, selecting the candidate object corresponding to the second similarity greater than the preset fifth similarity threshold from the determined second similarity between the selected target object and each candidate object, and if the second similarity greater than the preset fifth similarity threshold is multiple, multiple candidate objects are selected; the preset fifth similarity threshold can be 0.7, 0.8, etc. Strategy three, sorting the second similarity between the target object and each candidate object, and then selecting the candidate object corresponding to the preset number of second similarities from large to small according to the sorting result; the preset number can be flexibly set according to needs, for example, 2, 4, 5, etc.

[0091] Embodiment 6:

[0092] In order to make the similarity between the target object and the candidate object more accurate, on the basis of each of the above embodiments, in the embodiment of the present application, the attribute information includes age information, gender information and skin color information.

[0093] The matching of the to-be-identified attribute information and the candidate attribute information of the candidate object, the correction of the first similarity according to the matching result, and the determination of the second similarity between the target object and the candidate object include:

[0094] For each to-be-identified attribute information, the to-be-identified attribute information is matched with the corresponding candidate attribute information of the candidate object in the feature library to determine the similarity change value corresponding to the to-be-identified attribute information;

[0095] According to the similarity change value corresponding to each to-be-identified attribute information and the weight value corresponding to each to-be-identified attribute information determined in advance, the similarity change value between the target object and the candidate object is determined.

[0096] According to the similarity change value, the first similarity is corrected to determine a second similarity between the target object and the candidate object.

[0097] In the embodiment of the present application, first, the to-be-identified gait feature information is matched with the candidate gait feature information of the candidate object in the feature library to determine a first similarity between the target object and the candidate object. Then, the to-be-identified attribute information is matched with the candidate attribute information of the candidate object in the feature library to determine a similarity change value between the target object and the candidate object. Wherein, the to-be-identified attribute information is the same as the candidate attribute information of the candidate object in the feature library, or the to-be-identified attribute information is uncertain, at this time, the similarity change value between the target object and the candidate object is 0, and the to-be-identified attribute information is different from the candidate attribute information of the candidate object in the feature library, at this time, the similarity change value between the target object and the candidate object is a negative value. If the to-be-identified attribute information is multiple, the similarity change value corresponding to each to-be-identified attribute information is determined respectively. Then, the first similarity and the similarity change value between the target object and the candidate object are added to obtain a second similarity between the target object and the candidate object.

[0098] In the embodiment of the present application, the attribute information includes age information, gender information and skin color information, for each to-be-identified attribute information, the to-be-identified attribute information is matched with the corresponding candidate attribute information of the candidate object in the feature library to determine the similarity change value corresponding to the to-be-identified attribute information. The weight value corresponding to each to-be-identified attribute information is determined in advance, wherein the weight value corresponding to each to-be-identified attribute information reflects the importance of each to-be-identified attribute information for identity recognition, which can be set according to the demand. Also, after the attribute extraction model is trained, the accuracy of the attribute extraction model for each to-be-identified attribute information is determined, and then a high weight value is assigned to the to-be-identified attribute information with high accuracy. The similarity change value corresponding to each to-be-identified attribute information and the weight value corresponding to each to-be-identified attribute information are weighted and summed to determine the similarity change value between the target object and the candidate object. According to the similarity change value, the first similarity is corrected to determine a second similarity between the target object and the candidate object.

[0099] In the embodiment of the present application, for each to-be-identified attribute information, the similarity change value corresponding to the to-be-identified attribute information is determined; and according to the similarity change value corresponding to each to-be-identified attribute information and the weight value corresponding to each to-be-identified attribute information, the similarity change value between the target object and the candidate object is determined. The determined similarity change value between the target object and the candidate object is more accurate.

[0100] Embodiment 7:

[0101] In order to improve the efficiency of identity recognition, on the basis of the above embodiments, in the embodiment of the present application, after determining the first similarity between the target object and the candidate object, before matching the to-be-identified attribute information with the candidate attribute information of the candidate object, the method further comprises:

[0102] If the first similarity is greater than a preset third similarity threshold, it is determined that the target object is the candidate object; if the first similarity is less than a preset fourth similarity threshold, it is determined that the target object is not the candidate object.

[0103] If the first similarity is between the preset third similarity threshold and the preset fourth similarity threshold, the step of matching the to-be-identified attribute information with the candidate attribute information of the candidate object is performed; wherein the preset third similarity threshold is greater than the preset fourth similarity threshold.

[0104] In the embodiment of the present application, the preset third similarity threshold is greater than the preset first similarity threshold and the preset second similarity threshold, and the preset fourth similarity threshold is less than the preset first similarity threshold and the preset second similarity threshold.

[0105] After matching the to-be-identified gait feature information with the candidate gait feature information of the candidate object in the feature library to determine the first similarity between the target object and the candidate object, first, the size relationship between the first similarity and the preset third similarity threshold and the preset fourth similarity threshold is judged. If the first similarity is greater than the preset third similarity threshold, at this time, the first similarity does not need to be corrected according to the to-be-identified attribute information, and the target object can be directly determined as the candidate object. If the first similarity is less than the preset fourth similarity threshold, at this time, the first similarity also does not need to be corrected according to the to-be-identified attribute information, and the target object can be directly determined as not the candidate object.

[0106] If the first similarity is between the preset third similarity threshold and the preset fourth similarity threshold, at this time, in order to make the determination of the identity of the target object more accurate, the to-be-identified attribute information is matched with the candidate attribute information of the candidate object in the feature library to determine the similarity change value between the target object and the candidate object; according to the first similarity between the target object and the candidate object and the similarity change value, the second similarity between the target object and the candidate object is determined. Then, whether the target object is the candidate object is determined according to the second similarity between the target object and the candidate object.

[0107] If the first similarity is greater than the third preset similarity threshold, the target object is directly determined as the candidate object in the embodiment of the present application. If the first similarity is less than the fourth preset similarity threshold, the target object is directly determined as not the candidate object. Therefore, the efficiency of identity recognition is improved under the premise of ensuring the accuracy of identity recognition.

[0108] Figure 2 An identity recognition schematic diagram provided by the embodiment of the present application is shown in FIG. 1. First, a pedestrian video is collected, which is divided into continuous RGB images according to frames. Pedestrian attribute information to be recognized is extracted, and pedestrian gait feature information to be recognized is extracted. Identity recognition is performed in combination with the attribute information to be recognized and the gait feature information to be recognized. Figure 2

[0109] The pedestrian video can be a pedestrian walking video captured by a monitoring camera, a vehicle event data recorder or the like. A video segment in which the pedestrian walks clearly and completely is intercepted from the pedestrian video, and the video segment is divided into a continuous RGB image sequence according to the order of frames.

[0110] An image with the best imaging quality is selected from the continuous RGB image sequence, and attribute information to be recognized of the pedestrian in the image is extracted. The attribute information includes the gender of the pedestrian, the age of the pedestrian and the skin color of the pedestrian. The gender of the pedestrian is divided into male and female. The extracted gender attribute information should be as follows: {male: confidence that the pedestrian is male; female: confidence that the pedestrian is female}. The age of the pedestrian is difficult to accurately determine, so the age range to which the pedestrian belongs can be considered. The extracted age attribute information should be as follows: {child: confidence that the pedestrian is a child; adult: confidence that the pedestrian is an adult but not an old person; old person: confidence that the pedestrian is an old person}. The skin color of the pedestrian is divided into yellow skin, black skin and white skin. The extracted skin color attribute information should be as follows: {yellow skin: confidence that the pedestrian has yellow skin; black skin: confidence that the pedestrian has black skin; white skin: confidence that the pedestrian has white skin}. The attribute information of the pedestrian is extracted by an attribute extraction model. The attribute extraction model includes a feature extraction stage and a discrimination stage. The feature extraction stage is mainly composed of a multi-layer convolutional neural network, which is used to extract attribute-related feature information. The discrimination stage mainly includes a full connection layer and a SoftMax layer, which discriminates and classifies according to the extracted feature information to obtain the attribute information of the pedestrian. The RGB image of the pedestrian is input into the attribute extraction model, and the attribute information to be recognized of the pedestrian is output. The attribute information of the pedestrian can be saved as a json file or a txt file.

[0111] ​The continuous RGB image sequence is processed, and the RGB image is segmented into gait contour images by using a background subtraction method or a pre-trained image segmentation model to obtain a gait contour image sequence of the pedestrian; then the gait contour image sequence is sent into a pre-trained gait feature extraction model to extract gait feature information of the pedestrian. The image segmentation model can use existing segmentation models such as U-Net, DeepLabv3, and PSPNet, or a self-designed and pre-trained segmentation model. The gait feature extraction model is composed of a multi-layer convolutional neural network, and a triplet loss function is used in training. The final output value of the loss function should be within a given range, which needs to be set according to the actual situation, and values such as 0.1 and 0.2 can be used.

[0112] After obtaining the to-be-identified attribute information and the to-be-identified gait feature information of the pedestrian by the above method, the candidate attribute information and the candidate gait feature information of each candidate object in the feature library can be obtained by the same method. First, the cosine distance between the to-be-identified gait feature information and the gait feature information of the candidate object in the feature library is calculated, and an initial similarity is obtained according to the cosine distance; then the to-be-identified attribute information and the candidate attribute information of the candidate object in the feature library are matched to determine a similarity change value. There are three cases: 1. The gender of the target object is the same as that of the candidate object, and the original similarity between them is kept unchanged; 2. The gender of the target object is different from that of the candidate object, and the initial similarity between them is appropriately reduced; 3. The gender of one of the target object and the candidate object is uncertain, and the initial similarity between them is kept unchanged. It should be noted that when the confidence of the male (or female) is higher than a given setting value, the pedestrian is considered to be male (or female); when the confidence of both the male and the female is not higher than the given setting value, the gender of the pedestrian is considered to be uncertain. The given setting value can be set according to the actual situation, and values such as 0.8 and 0.85 can be considered. According to the above method, the age attribute and the skin color attribute of the target object and the candidate object are compared again, and the reduction or non-reduction of the three comparisons uses an accumulation strategy. Specifically, the accumulation strategy is as follows:

[0113]

[0114] In the above formula, represents the required reduction when comparing attribute i, where i = 0, 1, 2, corresponding to the gender attribute, the age attribute, and the skin color attribute, respectively, and W i represents the weight value. Since the recognition accuracy of the pedestrian attribute extraction model for the gender attribute, the age attribute, and the skin color attribute may differ, we assign a larger weight value to the attribute with higher recognition accuracy. The specific weight values of the three attributes need to be set according to the specific recognition accuracy of the pedestrian attribute extraction model.ori represents the initial similarity, S min and S max respectively represent the lower limit and upper limit of the similarity involved in the calculation, if the initial similarity is lower than S min the target object and the candidate object are not the same person, and it is considered that the initial similarity is higher than S max the target object and the candidate object are the same person, which avoids the time-consuming of the algorithm being too high, S min and S max The specific values of S i_max represents the attribute category of the target object, g i_max represents the attribute category of the candidate object, c qi_max represents the confidence of the target object and the candidate object on the attribute category when q i_max is different from g i_max For example, in the calculation of the gender attribute, the confidence of the target object being male is 0.8, and the confidence of being female is 0.2, so q i_max represents male, the confidence of the candidate object being male is 0.2, and the confidence of being female is 0.8, so g i_max represents female, q i_max is different from g i_max , then The value of the candidate object is the confidence of being male 0.2.

[0115] According to the above formula, the deduction scores of the gender attribute, the age attribute and the skin color attribute are calculated respectively The three are added to the initial similarity, and the similarity of the target object and the candidate object is obtained. Then, according to the order from large to small, the identity information of the candidate object ranked first is considered as the identity information of the target object, and the recognition result is output.

[0116] Embodiment 8:

[0117] Figure 3 The identity recognition device structure schematic diagram provided by the embodiment of the application, the device comprises:

[0118] The first determination module 31 is used for determining the to-be-identified gait feature information and the to-be-identified attribute information of the target object in the to-be-processed video.

[0119] The second determination module 32 is used for matching the to-be-identified gait feature information with the candidate gait feature information of the candidate object, and determining the first similarity between the target object and the candidate object.

[0120] The third determining module 33 is configured to match the to-be-identified attribute information with candidate attribute information of the candidate object, correct the first similarity according to a matching result, and determine a second similarity between the target object and the candidate object.

[0121] The recognition module 34 is configured to determine whether the target object is the candidate object according to the second similarity.

[0122] The first determining module 31 is specifically configured to input each image in the to-be-processed video into a trained image segmentation model respectively, determine to-be-identified gait contour graphs of the target object in each image based on the image segmentation model, and determine a to-be-identified gait contour sequence according to each to-be-identified gait contour graph, wherein the image segmentation model is obtained by training first sample images labeled with gait contour position information; input the to-be-identified gait contour sequence into a trained gait feature extraction model, and determine to-be-identified gait feature information corresponding to the to-be-identified gait contour sequence based on the gait feature extraction model, wherein the gait feature extraction model is obtained by training sample gait contour sequences labeled with sample gait feature information.

[0123] The first determining module 31 is specifically configured to determine a to-be-identified image in the to-be-processed video, input the to-be-identified image into a trained attribute extraction model, and determine to-be-identified attribute information of the target object in the to-be-identified image based on the attribute extraction model, wherein the attribute extraction model is obtained by training second sample images labeled with sample object attribute information.

[0124] The recognition module 34 is specifically configured to determine that the target object is the candidate object if the second similarity is greater than a preset first similarity threshold.

[0125] The recognition module 34 is specifically configured to determine the second similarity between the target object and each candidate object, select a candidate object corresponding to a maximum value of the determined second similarities, and determine that the target object is the selected candidate object.

[0126] The recognition module 34 is specifically configured to select a candidate object corresponding to a maximum second similarity from second similarities greater than a preset second similarity threshold, and determine that the target object is the selected candidate object.

[0127] The third determining module 33 is specifically configured to, for each to-be-identified attribute information, match the to-be-identified attribute information with corresponding candidate attribute information of a candidate object in the feature library, determine a similarity change value corresponding to the to-be-identified attribute information, and determine a similarity change value of the target object and the candidate object according to the similarity change value corresponding to each to-be-identified attribute information and a weight value corresponding to each to-be-identified attribute information determined in advance; and correct the first similarity according to the similarity change value, and determine a second similarity of the target object and the candidate object.

[0128] The identification module 34 is further configured to, if the first similarity is greater than a third preset similarity threshold, determine that the target object is the candidate object, if the first similarity is less than a fourth preset similarity threshold, determine that the target object is not the candidate object, and if the first similarity is between the third preset similarity threshold and the fourth preset similarity threshold, trigger the third determining module; the third preset similarity threshold is greater than the fourth preset similarity threshold.

[0129] Embodiment 9:

[0130] On the basis of the above-mentioned embodiments, the embodiment of the present application further provides an electronic device, as shown in the accompanying drawings, comprising a processor 301, a communication interface 302, a memory 303 and a communication bus 304, wherein the processor 301, the communication interface 302 and the memory 303 complete mutual communication through the communication bus 304. Figure 4

[0131] The memory 303 stores a computer program, and when the program is executed by the processor 301, the processor 301 executes the following steps:

[0132] Determine to-be-identified gait feature information and to-be-identified attribute information of a target object in a to-be-processed video;

[0133] Match the to-be-identified gait feature information with candidate gait feature information of a candidate object, and determine a first similarity of the target object and the candidate object;

[0134] Match the to-be-identified attribute information with candidate attribute information of the candidate object, correct the first similarity according to a matching result, and determine a second similarity of the target object and the candidate object;

[0135] Determine whether the target object is the candidate object according to the second similarity.

[0136] ​Based on the same inventive concept, the embodiment of the present application further provides an electronic device. Since the principle of solving problems of the electronic device is similar to that of the identity recognition method, implementation of the electronic device can be referred to implementation of the method, and details are not described herein.

[0137] The electronic device provided by the embodiment of the present application can be a desktop computer, a portable computer, a smart phone, a tablet computer, a personal digital assistant (PDA), a network side device, and the like.

[0138] The communication bus of the electronic device can be a peripheral component interconnect (PCI) bus, an extended industry standard architecture (EISA) bus, or the like. The communication bus can be divided into an address bus, a data bus, a control bus, and the like. For the convenience of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0139] The communication interface 302 is used for communication between the electronic device and other devices.

[0140] The memory can include a random access memory (RAM) and can also include a non-volatile memory (NVM), for example, at least one disk memory. Optionally, the memory can also be at least one storage device located away from the aforementioned processor.

[0141] The processor can be a general-purpose processor, including a central processing unit, a network processor (NP), and the like; can also be a digital signal processor (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and the like.

[0142] In the embodiment of the present application, when the processor executes the program stored on the memory, the gait feature information to be recognized and the attribute information to be recognized of the target object in the video to be processed are determined; the gait feature information to be recognized is matched with the candidate gait feature information of the candidate object to determine the first similarity between the target object and the candidate object; the attribute information to be recognized is matched with the candidate attribute information of the candidate object, and the first similarity is corrected according to the matching result to determine the second similarity between the target object and the candidate object; and whether the target object is the candidate object is determined according to the second similarity. In the embodiment of the present application, the gait feature information to be recognized and the attribute information to be recognized of the target object in the video to be processed are determined, the first similarity between the target object and the candidate object is determined based on the gait feature information to be recognized, the first similarity between the target object and the candidate object is corrected based on the attribute information to be recognized to obtain the second similarity, and then the identity recognition is performed according to the second similarity. In the embodiment of the present application, the first similarity between the target object and the candidate object determined based on the gait feature information to be recognized is corrected by using the attribute information to be recognized, whether the target object is the candidate object is determined according to the second similarity obtained after the correction, and the accuracy of the identity recognition of the target object is improved.

[0143] Embodiment 10:

[0144] On the basis of the above-mentioned embodiments, the embodiment of the present application further provides a computer storage readable storage medium, wherein the computer readable storage medium stores a computer program executable by an electronic device, and when the program runs on the electronic device, the electronic device is caused to perform the following steps:

[0145] determining the gait feature information to be recognized and the attribute information to be recognized of the target object in the video to be processed;

[0146] matching the gait feature information to be recognized with the candidate gait feature information of the candidate object to determine the first similarity between the target object and the candidate object;

[0147] matching the attribute information to be recognized with the candidate attribute information of the candidate object, correcting the first similarity according to the matching result to determine the second similarity between the target object and the candidate object;

[0148] determining whether the target object is the candidate object according to the second similarity.

[0149] Based on the same inventive concept, the embodiment of the present application also provides a computer readable storage medium, since the principle of solving problems of the processor in executing the computer program stored on the computer readable storage medium is similar to the identity recognition method, therefore, the implementation of the processor in executing the computer program stored on the computer readable storage medium can be referred to the implementation of the method, and the repeated parts will not be described herein.

[0150] The computer readable storage medium can be any available medium or data storage device that the processor in the electronic device can access, including but not limited to magnetic memories such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc., optical memories such as CDs, DVDs, BD, HVD, etc., and semiconductor memories such as ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid state disk (SSD), etc.

[0151] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0152] These computer program instructions can also be stored in a computer readable memory that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable memory produce a product including instruction apparatus, which implements the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0153] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer implemented process, so that the instructions executed on the computer or other programmable device provide a product for implementing the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0154] While the preferred embodiments of the application have been described, additional variations and modifications can be made to these embodiments by those skilled in the art once they have the benefit of the present disclosure without departing from the spirit and scope of the application. Accordingly, it is intended that the appended claims include all such modifications and variations as fall within the scope of the present application.

[0155] It is apparent that those skilled in the art can make various changes and modifications to the application without departing from the spirit and scope of the application. It is therefore intended that the present application cover all such changes and modifications that are within its scope.

Claims

1. An identity recognition method, characterized in that, The method includes: Determine the gait features and attribute information to be identified for the target object in the video to be processed; Specifically, in the video to be processed, an image to be identified is determined, and the image to be identified is input into a trained attribute extraction model. Based on the attribute extraction model, the attribute information of the target object in the image to be identified is determined. The attribute extraction model is trained using a second sample image labeled with the attribute information of the sample object. The gait feature information to be identified is matched with the candidate gait feature information of the candidate object to determine the first similarity between the target object and the candidate object; For each attribute information to be identified, the attribute information to be identified is matched with the corresponding candidate attribute information of the candidate object in the feature library to determine the similarity change value corresponding to the attribute information to be identified; based on the similarity change value corresponding to each attribute information to be identified and the pre-determined weight value corresponding to each attribute information to be identified, the similarity change value between the target object and the candidate object is determined; the first similarity is corrected based on the similarity change value to determine the second similarity between the target object and the candidate object; After the attribute extraction model is trained, the accuracy of the attribute extraction model for each type of attribute information to be identified is determined, and high weight values ​​are assigned to the attribute information with high accuracy. Whether the target object is a candidate object is determined based on the second similarity. Among them, if the attribute information to be identified is the same as the candidate attribute information of the candidate object in the feature library, the similarity change value between the target object and the candidate object is 0; if the attribute information to be identified is different from the candidate attribute information of the candidate object in the feature library, the similarity change value between the target object and the candidate object is negative; the first similarity and the similarity change value between the target object and the candidate object are added together to obtain the second similarity between the target object and the candidate object.

2. The method as described in claim 1, characterized in that, The gait feature information to be identified for the target object in the video to be processed includes: Each image in the video to be processed is input into the trained image segmentation model. Based on the image segmentation model, the gait contour map of the target object in each image is determined. The gait contour sequence to be identified is determined based on each gait contour map. The image segmentation model is trained using a first sample image labeled with gait contour position information. The gait contour sequence to be identified is input into the trained gait feature extraction model, and the gait feature information corresponding to the gait contour sequence to be identified is determined based on the gait feature extraction model; wherein, the gait feature extraction model is obtained by training sample gait contour sequences labeled with sample gait feature information.

3. The method as described in claim 1, characterized in that, Determining whether the target object is the candidate object based on the second similarity includes: If the second similarity is determined to be greater than the preset first similarity threshold, then the target object is determined to be the candidate object.

4. The method as described in claim 1, characterized in that, The candidate objects include at least two, and determining whether the target object is a candidate object based on the second similarity includes: Determine the second similarity between the target object and each candidate object; select the candidate object corresponding to the maximum value of each determined second similarity; determine the target object as the selected candidate object.

5. The method as described in claim 4, characterized in that, The candidate object corresponding to the maximum value of each determined second similarity is selected; Determining the target object as a selected candidate object includes: From the second similarity scores that are greater than the preset second similarity threshold, select the candidate object corresponding to the highest second similarity score; The target object is determined to be the selected candidate object.

6. The method as described in claim 1, characterized in that, After determining the first similarity between the target object and the candidate object, and before matching the attribute information to be identified with the candidate attribute information of the candidate object, the method further includes: If the first similarity is greater than a preset third similarity threshold, the target object is determined to be the candidate object; if the first similarity is less than a preset fourth similarity threshold, the target object is determined not to be the candidate object. If the first similarity is between the preset third similarity threshold and the preset fourth similarity threshold, the step of matching the attribute information to be identified with the candidate attribute information of the candidate object is performed; wherein, the preset third similarity threshold is greater than the preset fourth similarity threshold.

7. An identity recognition device, characterized in that, The device includes: The first determining module is used to determine the gait feature information and attribute information to be identified of the target object in the video to be processed; Specifically, in the video to be processed, an image to be identified is determined, and the image to be identified is input into a trained attribute extraction model. Based on the attribute extraction model, the attribute information of the target object in the image to be identified is determined. The attribute extraction model is trained using a second sample image labeled with the attribute information of the sample object. The second determining module is used to match the gait feature information to be identified with the candidate gait feature information of the candidate object to determine the first similarity between the target object and the candidate object; The third determining module is used to match each attribute information to be identified with the corresponding candidate attribute information of the candidate object in the feature library to determine the similarity change value corresponding to the attribute information to be identified; determine the similarity change value between the target object and the candidate object based on the similarity change value corresponding to each attribute information to be identified and the pre-determined weight value corresponding to each attribute information to be identified; and correct the first similarity based on the similarity change value to determine the second similarity between the target object and the candidate object. After the attribute extraction model is trained, the accuracy of the attribute extraction model for each type of attribute information to be identified is determined, and high weight values ​​are assigned to the attribute information with high accuracy. The identification module is used to determine whether the target object is the candidate object based on the second similarity. Among them, if the attribute information to be identified is the same as the candidate attribute information of the candidate object in the feature library, the similarity change value between the target object and the candidate object is 0; if the attribute information to be identified is different from the candidate attribute information of the candidate object in the feature library, the similarity change value between the target object and the candidate object is negative; the first similarity and the similarity change value between the target object and the candidate object are added together to obtain the second similarity between the target object and the candidate object.

8. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1-6.

Images