User identity recognition method and device

Abstract

The application provides a user identity recognition method and device, the user identity recognition method comprises: obtaining face information and first body state information of a user; obtaining at least one user identity information associated with the face information and second body state information associated with the at least one user identity information based on the face information; and obtaining target identity information of the user based on the face information, the at least one user identity information, the first body state information and the second body state information. The method can calculate the similarity of the face information and the body state information of the user to be detected respectively, and comprehensively evaluate the real identity of the user by combining the face similarity and the body state similarity, so that the similarity of the face recognition is not required to be adjusted multiple times, and the accuracy of the user identity recognition is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of target detection, and in particular to a user identity recognition method and device. BACKGROUND

[0002] In the process of digital management of fitness venues, the identities of the people entering and leaving need to be recognized to ensure the safety of the people entering and leaving.

[0003] In the prior art, in places with high personnel mobility, facial recognition technology is usually used to quickly identify the identity of the user. However, the facial recognition technology is very dependent on the setting of the similarity threshold. If the similarity threshold is set too high, the false rejection rate will be too high, and if the similarity threshold is set too low, the false acceptance rate will be too high. When the similarity threshold cannot adapt to the recognition environment, the recognition accuracy of the facial recognition technology is low. SUMMARY

[0004] The present application provides a user identity recognition method and device to solve the problem of difficulty in setting the similarity threshold when using facial recognition to quickly identify the identity of the user in the prior art, which results in low recognition accuracy of facial recognition technology, and improves the accuracy of user identity recognition.

[0005] The present application provides a user identity recognition method, comprising:

[0006] Obtaining facial information and first body posture information of a user, the first body posture information being used to represent the body shape or posture of the user;

[0007] Based on the facial information, obtaining at least one user identity information associated with the facial information and second body posture information associated with the at least one user identity information;

[0008] Based on the facial information, the at least one user identity information, the first body posture information and the second body posture information, obtaining target identity information of the user, the target identity information being one of the at least one user identity information.

[0009] According to the user identity recognition method provided by the present application, the at least one user identity information associated with the facial information and the second body posture information associated with the at least one user identity information are obtained based on the facial information, comprising:

[0010] Sending the facial information to a cloud server, the cloud server storing a plurality of user identity information and second body posture information associated with the plurality of user identity information;

[0011] receive the at least one user identity information and the second body state information associated with the at least one user identity information sent by the cloud server, the at least one user identity information being the identity information associated with the face information in the plurality of user identity information. According to the user identity recognition method provided by the application, the first body state information includes first human body features and first posture features, the second body state information includes second human body features and second posture features, and the target identity information of the user is obtained based on the face information, the at least one user identity information, the first body state information and the second body state information, including:

[0012] determining a face similarity corresponding to the at least one user identity information based on the face information and the at least one user identity information, the face similarity being used to represent the similarity between the face information and the face information in the at least one user identity information;

[0013] determining a human body similarity corresponding to the at least one user identity information based on the first human body features and the second human body features, the human body similarity being used to represent the similarity between the body shape information corresponding to the first human body features and the body shape information corresponding to the second human body features;

[0014] determining a posture similarity corresponding to the at least one user identity information based on the first posture features and the second posture features, the posture similarity being used to represent the similarity between the posture information corresponding to the first posture features and the posture information corresponding to the second posture features;

[0015] obtaining a target similarity corresponding to the at least one user identity information based on the face similarity, the human body similarity and the posture similarity;

[0016] determining the target identity information corresponding to the maximum value in the target similarity based on the target similarity corresponding to the at least one user identity information.

[0017] According to the user identity recognition method provided by the application, the first human body features include at least two, the second human body features are of the same type as the first human body features, and the human body similarity corresponding to the at least one user identity information is determined based on the first human body features and the second human body features, including:

[0018] applying the following formula to determine the human body similarity:

[0019]

[0020] wherein, the ​is the human body similarity; N is the number of the first human body features, N is a natural number greater than 1; is a feature value of a first feature in the at least two first human body features; is a feature value of an i-th feature in the at least two first human body features; is a feature value of a first feature in the at least two second human body features; is a feature value of an i-th feature in the at least two second human body features.

[0021] According to the user identity recognition method provided by the application, the first posture feature includes at least two, the second posture feature is of the same type as the first posture feature, and the posture similarity corresponding to the at least one user identity information is determined based on the first posture feature and the second posture feature, including:

[0022] The posture similarity is determined by using the following formula:

[0023] ,

[0024] Wherein, is the posture similarity; M is the number of the first posture features, M is a natural number greater than 0; is a maximum value of a feature value of an i-th feature in the at least two first posture features; is a maximum value of a feature value of an i-th feature in the at least two second posture features; is a minimum value of a feature value of an i-th feature in the at least two first posture features; is a minimum value of a feature value of an i-th feature in the at least two second posture features; is an average value of a feature value of an i-th feature in the at least two first posture features, is an average value of a feature value of an i-th feature in the at least two second posture features.

[0025] According to the user identity recognition method provided by the application, the target similarity corresponding to the at least one user identity information is obtained based on the face similarity, the human body similarity and the posture similarity, including:

[0026] The target similarity is determined by using the following formula:

[0027] ,

[0028] Wherein, is the target similarity; is the face similarity, a weight of the face similarity; a weight of the human body similarity, a weight of the human body similarity; a weight of the posture similarity, a weight of the posture similarity.

[0029] According to the user identity recognition method provided by the application, the first posture feature includes at least one of high and low shoulders, pelvic tilt, X-shaped legs and O-shaped legs.

[0030] The application further provides a user identity recognition device, comprising:

[0031] The acquisition module is configured to acquire face information and first body posture information of a user.

[0032] The first processing module is configured to obtain at least one user identity information associated with the face information and second body posture information corresponding to the at least one user identity information based on the face information.

[0033] The second processing module is configured to obtain target identity information of the user based on the face information, the at least one user identity information, the first body posture information and the second body posture information, wherein the target identity information is one of the at least one user identity information.

[0034] The application further provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the user identity recognition method according to any one of the above embodiments when executing the program.

[0035] The application further provides a non-transitory computer readable storage medium, which stores a computer program, wherein the computer program is executable on a processor to implement the user identity recognition method according to any one of the above embodiments.

[0036] The application further provides a computer program product, comprising a computer program, wherein the computer program is executable on a processor to implement the user identity recognition method according to any one of the above embodiments.

[0037] The user identity recognition method and device provided by the application can calculate the similarity of the face information and the body posture information of a user to be detected respectively, and comprehensively evaluate the real identity of the user by combining the face similarity and the body posture similarity of the user, thereby improving the accuracy of user identity recognition without adjusting the similarity of face recognition multiple times. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to make the technical solutions in the present application or prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort.

[0039] Figure 1 is one of the flowcharts of the user identity recognition method provided by the present application;

[0040] Figure 2 is another flowchart of the user identity recognition method provided by the present application;

[0041] Figure 3 is an interaction diagram of the user identity recognition method provided by the present application;

[0042] Figure 4 is a human joint distribution diagram provided by the present application;

[0043] Figure 5 is a structural diagram of the user identity recognition device provided by the present application;

[0044] Figure 6 is a structural diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0045] In order to make the technical solutions in the present application or prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort.

[0046] The user identity recognition method and device of the present application will be described below. Figures 1-5

[0047] Figure 1 is one of the flowcharts of the user identity recognition method provided by the present application, as shown in Figure 1 the user identity recognition method comprises the following steps:

[0048] Step 110, obtaining face information and first body shape information of a user, the first body shape information being used to represent the body shape or posture of the user.

[0049] In this step, the face information of the user includes the facial features of the user.

[0050] ​In this embodiment, the user facial information can be collected by an image collection device, for example, a camera is arranged at the entrance of a pedestrian passageway, which can collect facial information and body posture information of people passing by in real time under the current scene, form a video stream, and send the video stream to an AI (Artificial Intelligence) edge computing module.

[0051] In this embodiment, the AI edge computing module can perform operations such as calculation, analysis and storage on the sent video stream data.

[0052] For example, the AI edge computing module is provided with an AI face recognition algorithm and an AI body recognition algorithm, the AI face recognition algorithm can recognize the facial information to obtain a preliminary recognition result, and the AI body recognition algorithm can extract human proportion features from the body shape information in the first body posture information, or can extract posture features from the posture information.

[0053] In this embodiment, the AI edge computing module can save the received facial information and first body posture information, and save the feature information processed from the facial information and first body posture information.

[0054] Step 120, based on the facial information, obtaining at least one user identity information associated with the facial information and second body posture information associated with the at least one user identity information.

[0055] In this step, the second body posture information is of the same type as the first body posture information, and is used for similarity analysis with the first body posture information.

[0056] In this embodiment, the at least one user identity information and the corresponding second body data can be saved locally, or can be saved in other servers or terminals, and sent to the local for similarity calculation by other servers or terminals.

[0057] For example, the at least one user identity information and the corresponding second body information are saved in the AI edge computing module, and after the AI edge computing module obtains the facial information sent by the image collection device, the user information and the second body information saved locally are used to determine a plurality of user identity information associated with the facial information and the second body information of the corresponding user.

[0058] It should be noted that the user identity information stored in the AI edge computing module is pre-recorded, and within a certain time range, the recognition algorithm is easily affected by the appearance change of the user body shape (getting fat or getting thin) or hairstyle, and often cannot accurately identify the real identity of the user, but identifies other users with the closest appearance to the current user to be identified in the stored user identity information as the recognition object, which will cause misrecognition.

[0059] In this embodiment, the cloud server receives the face information of the user to be detected, and then performs similarity matching according to all the user images saved locally and preset conditions, and the corresponding matching results are user A, user B and user C.

[0060] In this embodiment, the second body information is completely same in type and quantity with the first body information.

[0061] In step 130, the target identity information of the user is obtained based on the face information, the at least one user identity information, the first body information and the second body information, and the target identity information is one of the at least one user identity information.

[0062] In this step, the face similarity of the user to be detected can be obtained by performing similarity matching on the real-time collected face information and the multiple identity information sent by the cloud server, and the body similarity of the user to be detected can be obtained by performing similarity matching on the real-time collected first body information and the multiple identity information sent by the cloud server and corresponding to the second body information.

[0063] It should be noted that the multiple face matching results obtained by performing similarity matching on the real-time collected face information and the multiple identity information sent by the cloud server will have a close similarity value, and at this time, the real identity of the user to be detected cannot be directly determined according to the face matching result.

[0064] In this embodiment, the identity information of the user to be detected can be comprehensively evaluated in combination with the face similarity and the body similarity.

[0065] In this embodiment, after the face similarity calculation is performed on the identity information of user A, user B and user C respectively by using the face information of the user to be detected A, the scores of the three are 0.9, 0.9 and 0.8 respectively, and after the similarity calculation is performed on the second body information of user A, user B and user C respectively by using the first body information of the user to be detected A, the scores are 0.8, 0.5 and 0.7 respectively, and then the identity information corresponding to the maximum value of the sum of the face similarity calculation score and the body similarity calculation score can be taken as the real identity information of the user to be detected A.

[0066] The embodiment of the present application provides a user recognition method, which performs similarity calculation on the face information and the body information of the user to be detected respectively, and comprehensively evaluates the real identity of the user by combining the face similarity and the body similarity, so that the similarity threshold of the face recognition does not need to be adjusted for many times, and the accuracy of the user identity recognition is improved.

[0067] In some embodiments, based on the face information, the at least one user identity information associated with the face information and the second body state information associated with the at least one user identity information are obtained by: sending the face information to a cloud server, the cloud server storing a plurality of user identity information and a plurality of user identity information associated second body state information; receiving at least one user identity information and at least one user identity information associated second body state information sent by the cloud server, the at least one user identity information being the identity information associated with the face information in the plurality of user identity information.

[0068] In this embodiment, the at least one user identity information and the corresponding second body data are stored in the cloud server and sent to the AI edge computing module for similarity calculation.

[0069] In this embodiment, the cloud server stores the identity information of a plurality of users, for example, the cloud server corresponding to the gym can store the identity information, face photo or validity period of the members and the like.

[0070] In this embodiment, the cloud server also stores the body shape data or posture data of each user at historical time.

[0071] It should be noted that since the user identity information stored in the cloud server is entered in advance, within a certain time range, the user in the face recognition, the recognition algorithm is easily affected by the appearance change of the user body shape (getting fat or getting thin) or hairstyle, and often cannot accurately identify the real identity of the user, but other users with the closest appearance to the current user to be identified in the stored user identity information are taken as the recognition object, which will cause misrecognition.

[0072] In this embodiment, the cloud server can also receive the face information sent by the AI edge computing module, and search for the user identity information matching the received face information from the user identity information stored in the cloud server.

[0073] In this embodiment, the cloud server receives the face information of user A at the current time, and the cloud server performs similarity matching according to the pre-set condition based on all the historical user images stored, and the corresponding matching results are user A, user B and user C.

[0074] In this embodiment, after the cloud server determines that the objects matching the face information of user A include user A, user B and user C according to the pre-set matching condition, the identity information and the second body state information of the three users are sent to the AI edge computing module.

[0075] Figure 2 is the second flowchart of the user identity recognition method provided by the application, in which Figure 2In the embodiment shown, the video acquisition subsystem of the fitness center acquires video stream data and sends it to the AI edge computing subsystem, the AI edge computing subsystem extracts the faces and body data (corresponding to the user's facial information and first body information) of the user to be detected from the video stream data, and calculates the body, body, or action characteristic values of the user, the AI edge computing subsystem sends the face data to the cloud server subsystem for personnel identity recognition, the cloud server subsystem compares with the existing user photos and preliminarily locates the user's identity information, and sends possible user information (such as at least one user identity information and corresponding second body information) to the AI edge computing subsystem, the AI edge computing subsystem compares the characteristic values of the body, body, and action of the possible users, and selects the user information corresponding to the highest comprehensive score of the face, body, body, and action characteristic values as the identity information of the user to be detected.

[0076] In some embodiments, the user identification method can be applied to the AI edge computing subsystem, which realizes accurate identification of user identity information by receiving real-time monitoring user information and cloud server sent user matching information.

[0077] Figure 3 is the interaction diagram of the user identity recognition method provided by the application, in Figure 3 In the embodiment shown, the image acquisition module in the video data acquisition subsystem acquires the images of the personnel entering and leaving the venue to form a video stream, and sends the video stream to the AI edge computing subsystem through the data transmission module to acquire the facial information of the user, the data transmission module in the AI edge computing subsystem is used to accept video stream data and send the personnel identity information data (corresponding to the facial information) recognized, the data storage module is used to store the video stream data to be processed and the personnel identity information data after processing, and the data processing module is used to process the video stream data into facial, body proportion, and posture characteristic values through AI face recognition algorithm and AI body recognition algorithm, and cooperate with the cloud service subsystem to confirm personnel identity information; the data transmission module in the cloud service subsystem is used to accept structured data and send personnel information recognition result data, the data storage module is used to store member basic information, facial photo, and validity period data, and the data processing module is used to compare member photos and identify member identity.

[0078] The user identification method provided by the embodiment of the application stores the identity information of all users and the corresponding second body information in the cloud server, thereby reducing the storage pressure of the local server and improving the stability of the server operation.

[0079] In some embodiments, the first body state information includes first human body features and first posture features, and the second body state information includes second human body features and second posture features. Based on the face information, the at least one user identity information, the first body state information, and the second body state information, the target identity information of the user is obtained, including: based on the face information and the at least one user identity information, determining a face similarity corresponding to the at least one user identity information, the face similarity being used to represent a similarity degree of the face information to the face information in the at least one identity information; based on the first human body features and the second human body features, determining a human body similarity corresponding to the at least one user identity information, the human body similarity being used to represent a similarity degree of the body shape information corresponding to the first human body features to the body shape information corresponding to the second human body features; based on the first posture features and the second posture features, determining a posture similarity corresponding to the at least one user identity information, the posture similarity being used to represent a similarity degree of the posture information corresponding to the first posture features to the posture information corresponding to the second posture features; based on the face similarity, the human body similarity, and the posture similarity, obtaining a target similarity corresponding to the at least one user identity information; and based on the target similarity corresponding to the at least one user identity information, determining an identity information corresponding to a maximum value in the target similarity as the target identity information.

[0080] In this embodiment, the first body state information can include a plurality of first human body features or first posture features of the user, for example, the first human body features can be shoulder width, upper body length, arm length, hip width, or lower limb length, etc.

[0081] Figure 4 is a schematic diagram of human body joint distribution provided by the present application, in Figure 4 In the embodiment shown in the figure, Figure 4 includes a plurality of human body joints, the length of the line connecting joint 3 and joint 6 represents shoulder width, the length of the line connecting joint 2 and joint 9 represents upper body length; the length of the line connecting joint 3 and joint 4 represents left upper arm length, the length of the line connecting joint 4 and joint 5 represents left lower arm length; the length of the line connecting joint 6 and joint 7 represents right upper arm length, the length of the line connecting joint 7 and joint 8 represents right lower arm length, the length of the line connecting joint 10 and joint 13 represents hip width, the length of the line connecting joint 10 and joint 11 represents left thigh length, the length of the line connecting joint 11 and joint 12 represents left lower leg length, the length of the line connecting joint 13 and joint 14 represents right thigh length, and the length of the line connecting joint 14 and joint 15 represents right lower leg length.

[0082] In this embodiment, the AI edge computing module can extract human body features corresponding to different joint lines from the obtained video stream, i.e., the first human body features include the above-mentioned 11 human body features.

[0083] In this embodiment, the first human body feature is obtained by collecting a user image in real time and extracting the human body feature from the user image; and the second human body feature is of the same type as the first human body feature, but is feature data obtained in advance and stored in a cloud server.

[0084] In this embodiment, since the feature values of the human body features in different video frames are not the same, the feature values of each type of human body feature in multiple video frames can be averaged, and compared and analyzed with the second feature values stored in the cloud server.

[0085] In this embodiment, the first posture feature can include a neck-shoulder angle, a hip angle, a left leg angle, and a right leg angle.

[0086] In this embodiment, the first posture feature is obtained by collecting a user image in real time and extracting the posture feature from the user image; and the second posture feature is of the same type as the first posture feature, but is feature data obtained in advance and stored in a cloud server.

[0087] In Figure 4 In the embodiment shown, the angle formed by the line connecting the joint node 1 and the joint node 2 and the line connecting the joint node 3 and the joint node 6 can represent the neck-shoulder angle; the angle formed by the line connecting the joint node 2 and the joint node 9 and the line connecting the joint node 10 and the joint node 13 can represent the hip angle; the angle formed by the line connecting the joint node 10 and the joint node 11 and the line connecting the joint node 11 and the joint node 12 can represent the left leg angle; and the angle formed by the line connecting the joint node 13 and the joint node 14 and the line connecting the joint node 14 and the joint node 15 can represent the right leg angle.

[0088] In this embodiment, the target similarity can include a face similarity, a human body similarity, and a posture similarity, which are calculated by weighting.

[0089] In this embodiment, the face similarity of the user can be determined by analyzing the face information and the user identity information returned by the cloud server through a face recognition algorithm.

[0090] In this embodiment, the face recognition algorithm can include an Eigenface method, a Local Binary Patterns (LBP) detection method, or a Fisherfaces detection algorithm.

[0091] In this embodiment, after the AI edge computing module receives the identity information of the user A, the user B and the user C, and the first human body features and the first posture features corresponding to the identity information, the face similarity, the human body similarity and the posture similarity between the to-be-detected user and the user A, the user B and the user C are calculated respectively, the three kinds of similarities corresponding to the user A, the user B and the user C are scored respectively, the target similarity is obtained, and the user identity information corresponding to the maximum target similarity is taken as the real identity information of the to-be-detected user.

[0092] The embodiment of the present application provides a user identification method, which compares and analyzes the real-time acquired face information, the first human body features and the first posture features of the to-be-detected user with the user identity information, the second human body features and the second posture features stored in the cloud server respectively, and comprehensively analyzes the acquired face similarity, the human body similarity and the posture similarity to obtain a target similarity, and then determines the identity information corresponding to the maximum target similarity as the real identity of the to-be-detected user, which can screen the identity of the user by using multiple features, does not need to increase additional user identity detection equipment, reduces the cost, and improves the identification accuracy of the user identity.

[0093] In some embodiments, the first human body features include at least two, the second human body features are of the same type as the first human body features, and the human body similarity corresponding to the at least one user identity information is determined based on the first human body features and the second human body features, including: applying the following formula to determine the human body similarity:

[0094]

[0095] wherein, the human body similarity is N is a natural number greater than 1; is the feature value of the first feature in the at least two first human body features; is the feature value of the i th feature in the at least two first human body features; is the feature value of the first feature in the at least two second human body features; is the feature value of the i th feature in the at least two second human body features.

[0096] In this embodiment, the first human body features include the shoulder width, the upper body length, the left upper arm length, the left lower arm length, the right upper arm length, the right lower arm length, the hip width, the left thigh length, the left calf length, the right thigh length and the right calf length, a total of 11 human body features, and the corresponding feature values are average values obtained according to the real-time collected user image data. The corresponding second human body features also include the above-mentioned 11 human body features, and the feature values of the corresponding second human body features are stored in the cloud server storage.

[0097] ​​In this embodiment, N is 11, the feature values corresponding to the 11 first human features are input into the above formula, and the similarities between the to-be-detected user and the second human features of the user A, the user B and the user C sent by the cloud server are respectively calculated to obtain the human similarities of the to-be-detected user and the user A, the user B and the user C respectively.

[0098] The embodiment of the present application provides a user identification method, which constructs a human similarity calculation formula, and combines 11 human features to comprehensively calculate the human similarities between a to-be-detected user and multiple identity information returned by a cloud server, realizes comprehensive analysis on multiple human features, and improves the reliability of calculating human similarity.

[0099] In some embodiments, the first posture feature includes at least two, the second posture feature is of the same type as the first posture feature, and the posture similarity corresponding to the at least one user identity information is determined based on the first posture feature and the second posture feature, including: applying the following formula to determine the posture similarity:

[0100] ,

[0101] Wherein, is the posture similarity; M is the number of the first posture feature, and M is a natural number greater than 0; is the maximum value of the feature value of the i-th feature in the at least two first posture features; is the maximum value of the feature value of the i-th feature in the at least two second posture features; is the minimum value of the feature value of the i-th feature in the at least two first posture features; is the minimum value of the feature value of the i-th feature in the at least two second posture features; is the average value of the feature value of the i-th feature in the at least two first posture features, is the average value of the feature value of the i-th feature in the at least two second posture features.

[0102] In this embodiment, the first posture feature includes the neck-shoulder angle, the hip angle, the left leg angle and the right leg angle of the user, a total of 4 posture features, and the corresponding feature values are obtained according to the real-time collected user image data. The corresponding second posture feature also includes the above-mentioned 4 posture features, and the feature values of the corresponding second posture feature are stored in the cloud server storage.

[0103] In this embodiment, M is 4, the feature values corresponding to the 4 first posture features are input into the above formula, and the similarities between the to-be-detected user and the second posture features of the user A, the user B and the user C sent by the cloud server are respectively calculated to obtain the posture similarities of the to-be-detected user and the user A, the user B and the user C respectively.

[0104] The embodiment of the present application provides a user identification method, which realizes comprehensive analysis of multiple posture features by constructing a posture similarity calculation formula and combining four posture features to comprehensively calculate posture similarity between a to-be-detected user and multiple identity information returned by a cloud server, and improves the reliability of calculating the posture similarity.

[0105] In some embodiments, obtaining a target similarity corresponding to at least one user identity information based on the face similarity, the body similarity and the posture similarity comprises: applying the following formula to determine the target similarity:

[0106] ,

[0107] wherein, the target similarity; the face similarity, a weight of the face similarity; the body similarity, a weight of the body similarity; the posture similarity, a weight of the posture similarity.

[0108] In this embodiment, the weight of the body similarity, the weight of the posture similarity and the weight of the face similarity can be set according to user demand.

[0109] In this embodiment, the sum of the weight of the body similarity, the weight of the posture similarity and the weight of the face similarity is 1, and accordingly, the weight of the face similarity can be set as 0.7, the weight of the body similarity can be set as 0.2, and the weight of the posture similarity can be set as 0.1.

[0110] In this embodiment, after obtaining the face similarity, the body similarity and the posture similarity between the to-be-detected user and user A, user B and user C sent by the cloud server, adjustable weights are set for the three similarities respectively to obtain target similarities between the to-be-detected user and user A, user B and user C respectively , and then the user identity information corresponding to the maximum value of the three target similarities is determined as the real identity information of the to-be-detected user, for example, after calculation, the target similarity value between the to-be-detected user and user A is the maximum, and the identity information of user A stored in the cloud server is the real identity information of the to-be-detected user.

[0111] The embodiment of the present application provides a user identification method, which determines target similarity corresponding to face similarity, body similarity and posture similarity through a formula, and takes the user identity information corresponding to the maximum target similarity as the real identity information of the user to be detected, so as to comprehensively evaluate the face features, body features and posture features, and further improve the accuracy of user identity identification.

[0112] In some embodiments, the first posture feature includes at least one of high-low shoulder, pelvic tilt, X-shaped leg and O-shaped leg.

[0113] It should be noted that the posture information of different users during exercise may be different. For example, a user presents high-low shoulder, pelvic tilt and the like when walking, or the lower limbs present X-shaped leg or O-shaped leg and the like. The posture information is used to identify users with different postures, assist the face recognition technology, and obtain accurate identification results.

[0114] In this embodiment, the high-low shoulder state of the user can be analyzed by acquiring the neck-shoulder angle data of the user, the pelvic tilt state of the user can be identified by acquiring the hip angle data of the user, and the X-shaped leg or O-shaped leg of the user can be identified by acquiring the left leg angle data and the right leg angle data of the user.

[0115] The embodiment of the present application provides a user identification method, which acquires a plurality of posture features corresponding to high-low shoulder, pelvic tilt, X-shaped leg and O-shaped leg, so as to identify different postures of the user, and further improve the accuracy of user identity identification.

[0116] The user identity identification device provided by the present application is described below, and the user identity identification device described below can be correspondingly referred to the user identity identification method described above.

[0117] Figure 5 is a structural schematic diagram of the user identity identification device provided by the present application, as Figure 5 shown, the present application further provides a user identity identification device, which comprises an acquisition module 510, a first processing module 520 and a second processing module 530.

[0118] The acquisition module 510 is used to acquire the face information and the first body information of the user.

[0119] The first processing module 520 is used to obtain at least one user identity information associated with the face information and second body information associated with the at least one user identity information based on the face information.

[0120] The second processing module 530 is used to obtain the target identity information of the user based on the face information, the at least one user identity information, the first body information and the second body information, and the target identity information is one of the at least one user identity information.

[0121] The embodiment of the present application provides a user identification device, which respectively performs similarity calculation on face information and body information of a user to be detected, and combines face similarity and body similarity of the user to comprehensively evaluate the real identity of the user, thereby improving the accuracy of user identity recognition without multiple adjustments of a similarity threshold of face recognition.

[0122] Figure 6 An example of a schematic diagram of a physical structure of an electronic device is shown in FIG. 1. Figure 6 As shown in FIG. 1, the electronic device can include a processor 610, a communications interface 620, a memory 630, and a communications bus 640, wherein the processor 610, the communications interface 620, and the memory 630 can communicate with each other through the communications bus 640. The processor 610 can invoke a logical instruction in the memory 630 to execute a user identity recognition method, which includes obtaining face information and first body information of a user; obtaining at least one user identity information associated with the face information based on the face information, and second body information associated with the at least one user identity information; and obtaining target identity information of the user based on the face information, the at least one user identity information, the first body information, and the second body information, wherein the target identity information is one of the at least one user identity information.

[0123] In addition, the logical instruction in the memory 630 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the present application essentially or partly, or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0124] In another aspect, the present application also provides a computer program product comprising a computer program, which can be stored on a non-transitory computer readable storage medium, and the computer program, when executed by a processor, enables a computer to perform the user identity identification method provided by the above-mentioned methods, and the method comprises: obtaining facial information and first body state information of a user; obtaining at least one user identity information associated with the facial information based on the facial information, and second body state information associated with the at least one user identity information; and obtaining target identity information of the user based on the facial information, the at least one user identity information, the first body state information and the second body state information, the target identity information being one of the at least one user identity information.

[0125] In another aspect, the present application also provides a computer program product comprising a computer program, which can be stored on a non-transitory computer readable storage medium, and the computer program, when executed by a processor, enables a computer to perform the user identity identification method provided by the above-mentioned methods, and the method comprises: obtaining facial information and first body state information of a user; obtaining at least one user identity information associated with the facial information based on the facial information, and second body state information associated with the at least one user identity information; and obtaining target identity information of the user based on the facial information, the at least one user identity information, the first body state information and the second body state information, the target identity information being one of the at least one user identity information.

[0126] The device embodiments described above are merely illustrative, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0127] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus necessary general hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0128] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; and although the present application has been described in detail with reference to the foregoing embodiments, it should be appreciated by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features thereof can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A user identification method, characterized in that, include: Acquire the user's facial information and first body posture information, wherein the first body posture information is used to represent the user's body shape or posture; Based on the facial information, at least one user identity information associated with the facial information and second body posture information associated with the at least one user identity information are obtained; Based on the facial information, the at least one user identity information, the first body posture information, and the second body posture information, the target identity information of the user is obtained, wherein the target identity information is one of the at least one user identity information; The first body posture information includes a first human body feature and a first posture feature; the second body posture information includes a second human body feature and a second posture feature; obtaining the user's target identity information based on the facial information, the at least one user identity information, the first body posture information, and the second body posture information includes: Based on the facial information and the at least one user identity information, a facial similarity corresponding to the at least one user identity information is determined, wherein the facial similarity is used to characterize the degree of similarity between the facial information and the facial information in the at least one user identity information; Based on the first human body feature and the second human body feature, the human body similarity corresponding to the at least one user identity information is determined, and the human body similarity is used to characterize the degree of similarity between the body shape information corresponding to the first human body feature and the body shape information corresponding to the second human body feature. Based on the first posture feature and the second posture feature, the posture similarity corresponding to the at least one user identity information is determined, and the posture similarity is used to characterize the degree of similarity between the posture information corresponding to the first posture feature and the posture information corresponding to the second posture feature. Based on the facial similarity, the human body similarity, and the pose similarity, the target similarity corresponding to the at least one user identity information is obtained; Based on the target similarity corresponding to the at least one user identity information, the identity information corresponding to the maximum value of the target similarity is determined as the target identity information; The first human body feature includes at least two features, and the second human body feature is of the same type as the first human body feature. The step of determining the human body similarity corresponding to the at least one user identity information based on the first human body feature and the second human body feature includes: The similarity of the human body is determined using the following formula: ， Among them, the The similarity of the human body is denoted as N; N is the number of the first human body features, where N is a natural number greater than 1. The feature value is the first feature among at least two first human body features; The feature value of the i-th feature among at least two first human body features; The feature value is the first feature among at least two second human body features; It is the feature value of the i-th feature among at least two second human body features.

2. The user identification method according to claim 1, characterized in that, The step of obtaining at least one user identity information associated with the facial information and second body posture information associated with the at least one user identity information based on the facial information includes: The facial information is sent to a cloud server, which stores multiple user identity information and second body posture information associated with the multiple user identity information; The system receives at least one user identity information and second body posture information associated with the at least one user identity information sent by the cloud server, wherein the at least one user identity information is the identity information associated with the facial information among the plurality of user identity information.

3. The user identification method according to claim 1, characterized in that, The first pose feature includes at least two features, and the second pose feature is of the same type as the first pose feature. Determining the pose similarity corresponding to the at least one user identity information based on the first pose feature and the second pose feature includes: The pose similarity is determined using the following formula: ， in, The pose similarity is M; M is the number of the first pose features, and M is a natural number greater than 0. It is the maximum value of the feature value of the i-th feature among at least two first pose features; It is the maximum value of the eigenvalues ​​of the i-th feature among at least two second pose features; It is the minimum value of the eigenvalues ​​of the i-th feature among at least two first pose features; It is the minimum value of the eigenvalues ​​of the i-th feature among at least two second pose features; The average of the feature values ​​of the i-th feature in at least two of the first pose features. It is the average of the feature values ​​of the i-th feature among at least two second pose features.

4. The user identification method according to claim 1, characterized in that, The step of obtaining the target similarity corresponding to the at least one user identity information based on the facial similarity, the human body similarity, and the pose similarity includes: The target similarity is determined using the following formula: ， in, The target similarity; The facial similarity is described above. The weights for the facial similarity; The similarity of the human body, The weights for the human body similarity; The pose similarity is... The weights for the pose similarity are denoted as .

5. The user identification method according to claim 1, characterized in that, The first postural feature includes at least one of uneven shoulders, pelvic tilt, knock-knees, and bowlegs.

6. A user identification device, employing the user identification method as described in claim 1, characterized in that, include: The acquisition module is used to acquire the user's facial information and initial body posture information; The first processing module is used to obtain, based on the facial information, at least one user identity information associated with the facial information and second body posture information associated with the at least one user identity information; The second processing module is used to obtain the target identity information of the user based on the facial information, the at least one user identity information, the first body posture information and the second body posture information, wherein the target identity information is one of the at least one user identity information.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the user identification method as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the user identification method as described in any one of claims 1 to 5.

Images