Teacher role identification method and electronic equipment

Abstract

The invention relates to the technical field of identity recognition, and discloses a teacher role recognition method and electronic equipment. According to the method, similarity comparison is carried out on teacher features identified in a current classroom and teacher features in a teacher feature group pre-stored in a database. If the similarity exceeds a preset threshold value, determining that the identified object is a teacher role; and if the similarity is lower than a threshold value, entering a dynamic replacement process, taking the feature with the highest occurrence frequency in the currently identified teacher features as new teacher feature information, and replacing the feature group with the lowest confidence coefficient in the teacher feature group. The process is carried out in each class, and the teacher feature group is updated step by step. According to the invention, the teacher features can be updated in real time, the possibility of misrecognition is reduced, and the recognition accuracy of teacher roles is improved; and along with the increase of the course sections of the teacher, the system can accumulate more feature data, the persistent information accumulation can more accurately match the teacher features, and the risk of misrecognition is reduced.

Description

Technical Field

[0001] This application relates to the technical field of identity recognition, and particularly to a method for recognizing teacher roles and an electronic device. Background Art

[0002] In the solutions for analyzing teacher behavior in teaching scenarios, the prerequisite is to first identify the role of the teacher. Related technologies are based on the statistical method of visual algorithms, using visual algorithms such as face recognition and person re-identification to identify the people appearing within the podium range and infer that they are teachers. However, this method is easily interfered with in complex scenarios and the recognition accuracy is insufficient. Summary of the Invention

[0003] An object of an embodiment of this application is to provide a method for recognizing teacher roles and an electronic device to solve the technical problem of insufficient recognition accuracy in the related technology when recognizing teacher roles.

[0004] To solve the above technical problem, a technical solution adopted in an embodiment of this application is: to provide a method for recognizing teacher roles, including: obtaining teacher data in the current classroom, and identifying the first teacher feature according to the teacher data; comparing the first teacher feature with each group of teacher features in the teacher feature group one by one, and calculating the similarity between the first teacher feature and each group of teacher features in the teacher feature group; the teacher feature group includes at least one group of teacher features and is stored in a preset database; if the similarity between the first teacher feature and the teacher feature in any group is greater than a preset threshold, it is determined that the object corresponding to the first teacher feature is a teacher role; if the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, enter the dynamic replacement process; where the dynamic replacement process includes: using the feature with the highest frequency of occurrence in the teacher features identified in the current classroom as the teacher feature information, and replacing the group of teacher features with the lowest confidence in the database with the teacher feature information to obtain an updated teacher feature group.

[0005] Among them, through the dynamic feature comparison and replacement mechanism, this method can update teacher features in real time, reduce the possibility of misidentification, and thus improve the recognition accuracy of teacher roles; as the number of teaching sessions of the teacher increases, more feature data can be accumulated, thereby forming a more comprehensive teacher feature model. This persistent information accumulation enables more accurate matching of teacher features during recognition and reduces the risk of misidentification; and over time, the feature changes of teachers can be dynamically analyzed to optimize the recognition algorithm and further improve the accuracy.

[0006] Optionally, the method further includes: initializing a teacher feature group; initializing the teacher feature group includes: triggering the start of the teaching analysis device based on the class schedule time, obtaining teacher data of the class through the started teaching analysis device, and obtaining target teacher features based on the teacher data, where the target teacher features are the face features and / or body features with the highest occurrence frequency in the teacher data; if the target teacher features are not included in the preset database, then use the target teacher features, the corresponding feature confidence, and the teacher information in the class schedule as the first group of teacher features, and store the first group of teacher features in the database; trigger the start of the teaching analysis device again based on the class schedule time, and obtain the teacher data of the class through the started teaching analysis device, and obtain the target teacher features again according to the teacher data; compare the target teacher features obtained again with the first group of teacher features, if the similarity is less than or equal to the preset threshold, then use the target teacher features obtained again, the corresponding feature confidence, and the teacher information in the class schedule as the second group of teacher features, and store the second group of teacher features in the database; at different time points when the teacher is in class, continuously start the teaching analysis device, obtain the target teacher features in the class, and compare the obtained target teacher features with the teacher features in the currently stored teacher feature group, if the similarity is less than or equal to the preset threshold, then use the newly obtained target teacher features, the corresponding feature confidence, and the teacher information in the class schedule as a new teacher feature group and store it in the database, and repeat the above process until the number of teacher feature groups reaches the preset number. Among them, this method can dynamically construct and optimize the teacher feature database to ensure the accuracy of teacher identification and the adaptability of the system. Through the triggering mechanism based on the class schedule time, the system can automatically collect and update teacher features, enabling it to adapt to the feature changes of teachers in different classroom scenarios. At the same time, the similarity comparison and dynamic storage mechanism can effectively reduce misidentification, improve the accuracy of feature matching, and ensure the accuracy of the teacher identity recognized in each class. As the number of classes increases, the system gradually accumulates multi-dimensional teacher features, making it more stable and reliable.

[0007] Optionally, when comparing the target teacher features obtained again with the first group of teacher features, if the similarity is greater than the preset threshold, the method further includes: determining the object corresponding to the target teacher features obtained again as the teacher role, and at the same time increasing the confidence corresponding to the first group of teacher features. Among them, by comparing the newly obtained features with the existing features, the teacher identity can be confirmed more accurately, reducing the possibility of misidentification; in addition, increasing the feature confidence enables dynamic adaptation to changes in the teacher identity, enhancing the reliability and applicability of the feature group.

[0008] Optionally, target teacher features are obtained based on teacher data. The target teacher features are the face features and / or body features with the highest occurrence frequency in the teacher data, including: obtaining face data and body data from the teacher data; obtaining face feature vectors from the face data and body feature vectors from the body data; the body feature vectors include human body outline features, clothing color features, and bone key point features; counting the occurrence times of each face feature vector and clustering similar face features to obtain the total frequency of each face feature cluster; counting the occurrence times of each body feature vector and clustering similar body features to obtain the total frequency of each body feature cluster; based on the total frequency of each face feature cluster and the total frequency of each body feature cluster, selecting the face feature vector and / or body feature vector with the most occurrences during the corresponding duration of the class as the target teacher features. This frequency statistics method can filter out accidentally occurring non-teacher features, ensuring that the extracted features are more reliable. Additionally, this method reduces the dependence on single-frame recognition results.

[0009] Optionally, the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to a preset threshold, including: obtaining the face features and / or body features with the highest occurrence frequency, as well as the face features and / or body features with the second and third highest occurrence frequencies from the first teacher feature; if the similarity between the face features and / or body features with the highest occurrence frequency and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, then comparing the similarity between the face features and / or body features with the second and third highest occurrence frequencies and each group of teacher features in the teacher feature group respectively; if the results of the similarity comparison are all less than or equal to the preset threshold, it is determined that the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold. By comparing features with different frequencies, the similarity of teacher features can be evaluated more comprehensively, ensuring the accuracy of recognition; and comparing frequency features layer by layer can effectively reduce the risk of misrecognition.

[0010] Optionally, if the similarity between the first teacher feature and the teacher feature in any group is greater than the preset threshold, it is determined that the object corresponding to the first teacher feature is a teacher role, including: obtaining the face features and / or body features with the highest occurrence frequency from the first teacher feature; comparing the similarity between the face features and / or body features with the highest occurrence frequency and the teacher features in the teacher feature group group by group. If the similarity between the teacher feature in any group is greater than the preset threshold, it is determined that the object corresponding to the face features and / or body features with the highest occurrence frequency in the first teacher feature is a teacher role. This method can effectively improve the accuracy and robustness of teacher identity recognition through a feature screening and similarity comparison mechanism based on frequency statistics.

[0011] Optionally, if the similarity between the first teacher feature and the teacher features in any group is greater than a preset threshold, the method further includes: increasing the corresponding confidence level of the group of teacher features whose similarity to the first teacher feature is greater than the preset threshold.

[0012] Optionally, if the similarity between the first teacher feature and the teacher features in any group is greater than a preset threshold, the method further includes: increasing the corresponding confidence level of the group of teacher features whose similarity to the first teacher feature is greater than the preset threshold. Among them, by increasing the confidence level of the group of teacher features with higher similarity, not only the trust level and recognition accuracy of the features are enhanced, but also the dynamic adaptation ability is supported.

[0013] Optionally, increasing the corresponding confidence level of the group of teacher features whose similarity to the first teacher feature is greater than a preset threshold includes: obtaining the current confidence level corresponding to the group of teacher features whose similarity to the most frequently occurring face feature and / or body feature is greater than the preset threshold; summing the current confidence level corresponding to the group of teacher features whose similarity is greater than the preset threshold and the confidence level corresponding to the first teacher feature to obtain a new confidence level; using the calculated new confidence level as the confidence level of the group of teacher features whose similarity is greater than the preset threshold and updating it. Among them, through the updated confidence level, the identity recognition is more accurate and the probability of misrecognition is reduced.

[0014] To solve the above technical problems, a technical solution adopted in an embodiment of the present application is: to provide an electronic device, including: a memory and a processor, the memory is connected to the processor, and the processor is configured to execute one or more computer programs stored in the memory. When the processor executes the one or more computer programs, the electronic device implements a teacher role recognition method applied to the electronic device.

[0015] To solve the above technical problems, a technical solution adopted in an embodiment of the present application is: to provide a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by an electronic device, the electronic device executes the above teacher role recognition method.

[0016] To solve the above technical problems, a technical solution adopted in an embodiment of the present application is: to provide a computer program product, the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by an electronic device, the electronic device executes the above teacher role recognition method.

[0017] The electronic device, the non-volatile computer-readable storage medium, and the computer program product have the beneficial effects corresponding to the above teacher role recognition method. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for the description of the embodiments of the present application. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 is an application scenario diagram of a teacher role recognition method provided by an embodiment of the present application;

[0020] Figure 2 is a flowchart of a teacher role recognition method provided by an embodiment of the present application;

[0021] Figure 3 is a flowchart of a method for initializing a teacher feature group provided by an embodiment of the present application;

[0022] Figure 4 is a flowchart of a method for determining that the object corresponding to the second teacher feature is a teacher role provided by an embodiment of the present application;

[0023] Figure 5 is a schematic structural diagram of a teacher role recognition device provided by an embodiment of the present application;

[0024] Figure 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0025] In order to make the objectives, technical solutions and advantages of the present application clearer, the following further details the present application in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts fall within the protection scope of the present application.

[0026] It should be noted that if there is no conflict, the various features in the embodiments of the present application can be combined with each other, and all are within the protection scope of the present application. In addition, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order from the module division in the device or the flowchart. Furthermore, the terms "first", "second", "third", etc. used in the present application do not limit the data and execution order, but only distinguish the same items or similar items with basically the same functions and effects.

[0027] In a teaching scenario, a solution for analyzing teachers' behaviors usually needs to first identify the role of the teacher. Because only by accurately identifying the teacher can the behaviors in the classroom be effectively tracked, analyzed, and evaluated. Teachers' behavior analysis usually relies on teacher role identification as the basis, and this process usually depends on various technical means for feature extraction and comparison. Once the teacher role is identified, behavior analysis can be carried out on this basis, including evaluations of teaching methods, classroom management, teacher-student interaction, etc., so as to provide data support and decision-making basis for the improvement of teaching quality.

[0028] When identifying the teacher role in related technologies, it is usually a statistical method based on visual algorithms, using computer vision technologies such as face recognition, person re-identification, and object detection to identify and infer the identities of the people within the podium range. In this process, first, images or video data of the classroom scenario are captured by a camera. Face recognition technology is used to detect and identify the facial features that may be the teacher, or person re-identification technology is used to identify and track based on personal appearance features (such as clothing, body shape, gait, etc.). Through these visual algorithms, the system can locate and confirm the identity of the person within the podium area, and thus infer whether the person is a teacher. However, this identification method based on visual algorithms faces great challenges in complex scenarios. For example, in a classroom environment, the teacher may be partially blocked by teaching aids, books, other students, or the podium, etc., which will seriously affect the accuracy of the recognition algorithm; even if the teacher stands on the podium, due to occlusion or side display, etc., the image lacks complete facial or body features, making it difficult for the visual algorithm to accurately identify. Another example is that in a classroom, there are usually multiple students and teachers moving together. Especially in interactive teaching, teachers and students may frequently enter and exit the podium range, resulting in a complex and dynamic variety of people in the scene. At this time, the visual algorithm may not be able to effectively distinguish between teachers and students. Especially in the case of unclear perspectives or multiple people overlapping, there are prone to misidentifications or missed identifications. Another example is that the classroom environment itself may be full of interference factors, such as the background in the classroom, windows, blackboards, projection screens, etc. These complex backgrounds will increase the difficulty of the recognition task. In some cases, the interference of the background may cause the recognition algorithm to misidentify other objects or environmental elements as teachers, affecting the final recognition result. Therefore, there is a technical problem of insufficient recognition accuracy when related technologies identify the teacher role.

[0029] To solve the above problems, the embodiments of the present application provide a method for identifying the teacher role. This method gradually improves the recognition accuracy through cross-class dynamic feature comparison and feature update mechanisms. The teacher features are identified and stored in the first class session, and in subsequent class sessions, the teacher feature information is continuously compared, updated, and replaced to accurately identify the teacher role. Specifically, the method for identifying the teacher role mainly includes the following key points:

[0030] Initialization and dynamic update of the teacher feature group: In the classroom, teacher feature information accumulates continuously as each class progresses, enhancing the richness and accuracy of the feature library.

[0031] Similarity comparison and confidence update: By comparing the teacher features extracted in the current class with the existing feature group, the feature library is gradually updated, thereby improving the recognition accuracy and reducing the risk of misrecognition.

[0032] Step-by-step matching across multiple class sessions: By accumulating the teacher's behavioral features over multiple class sessions (including consecutive classes or the entire semester), the model is gradually optimized to make the recognition of the teacher's role in a complex classroom environment more accurate.

[0033] The teacher role recognition method of the embodiments of this application can effectively filter out accidental interferences (such as students entering the podium area) through cross-class cumulative analysis, find commonalities in the feature information of different class sessions, and reduce the probabilities of misrecognition and missed recognition. As the number of class sessions the teacher has is increasing, the understanding of the teacher's features becomes deeper and the recognition accuracy gradually improves. Moreover, this method does not require the teacher to actively provide face information, avoiding the trouble of manual data uploading in traditional methods, thus improving the convenience of use and the acceptance of teachers. Also, by dynamically replacing the feature group with low confidence, it ensures that the teacher feature group always maintains the most effective and representative information, avoiding recognition errors caused by using outdated or incorrect features for a long time. Generally, this method effectively combines cross-class analysis and local storage of feature information, not only improving the recognition accuracy but also solving problems such as insufficient accuracy and cumbersome operations that may occur in traditional methods. As the number of class sessions the teacher has increases, it can more accurately recognize the teacher's role and has great application potential. Additionally, the teacher feature group for storing teacher features is pre-stored in the database, and the data does not rely on external cloud storage, thus ensuring data security; moreover, in the case of local storage, the database can be managed, updated, and optimized by itself to ensure the availability and security of the data. As the classroom data is updated, the data in the teacher feature group is then updated. Thus, through the dynamic update of the teacher feature group, the recognition accuracy can be continuously optimized and the generalization ability can be enhanced.

[0034] The application scenarios of the teacher role recognition method of the embodiments of this application can include intelligent classrooms or intelligent education environments, etc., and the application scope includes classroom monitoring and management systems in schools, training institutions, and online education platforms.

[0035] Optionally, refer to Figure 1 , Figure 1It is an application scenario diagram of a teacher role identification method provided by an embodiment of the present application, and the application scenario includes a video acquisition device, a computing device and a storage device. The video acquisition device is used to capture the video stream in the classroom in real time and obtain the dynamic information of the teacher and the students. The video acquisition device includes a camera, which is the core hardware device for teacher role identification. The image data collected by the camera provides basic data for the subsequent acquisition of facial features and human body features. A high-definition camera or a 360-degree panoramic camera can be used to ensure that the entire classroom area, especially the podium range and the full picture of the teacher's activities can be covered. In large classrooms, multiple cameras may be required for multi-angle and all-round monitoring. The camera can be installed on the ceiling, corner or wall of the classroom to ensure that the teacher's activities and features within the podium range can be clearly captured.

[0036] The computing device is used to process, analyze and calculate the video data collected by the camera, and execute the teacher role identification method of the embodiment of the present application, for example, to execute initialization of the teacher feature group, which includes at least one group of teacher features and is stored in a preset database; to obtain the first teacher feature identified in the current classroom; to compare the first teacher feature with each group of teacher features in the teacher feature group one by one, and to calculate the similarity between the first teacher feature and each group of teacher features in the teacher feature group; if the similarity between the first teacher feature and the teacher features in any group is greater than a preset threshold, then it is determined that the object corresponding to the first teacher feature is a teacher role; if the first teacher feature is similar to the teacher features in the teacher feature group If the similarity of each group of teacher features is less than or equal to a preset threshold, the dynamic replacement process is entered; wherein the dynamic replacement process includes: taking the most frequently appearing feature among the teacher features identified in the current class as the teacher feature information, and replacing the teacher feature information with a group of teacher features with the lowest confidence in the database to obtain an updated teacher feature group; in the next class, obtaining the second teacher features identified in the next class, and comparing the similarity of the second teacher features with the updated teacher feature group group by group; if the similarity is greater than the preset threshold, it is determined that the object corresponding to the second teacher feature is the teacher role; if the similarity is less than or equal to the preset threshold, the dynamic replacement process is executed.

[0037] The computing device can be a local server or an edge computing device with powerful computing capabilities. If needed, a distributed computing architecture can also be used to process multi-source video data.

[0038] The storage device can be a local storage device, and the teacher's characteristic data, videos and images during the teaching process, and other data can be stored in the local device. Since the data is stored locally and not uploaded to the cloud, high-security protection measures (such as encrypted storage, access control, etc.) can be configured to ensure data security.

[0039] The teacher role recognition method in the embodiments of the present application can be applied to the above application scenarios. Through the collaborative work of video acquisition devices, computing devices, and storage devices, the system is supported to perform teacher role recognition, feature update, and dynamic replacement in complex teaching environments, ensuring the accuracy of recognition results and the security of data. The above application scenarios are only used as an example and may also include other application scenarios.

[0040] See Figure 2 , Figure 2 is a flowchart of a teacher role recognition method provided by an embodiment of the present application. The method includes the following steps:

[0041] S11. Initialize a teacher feature group, which includes at least one set of teacher features and is stored in a preset database.

[0042] The teacher feature group refers to a set of feature collections used to identify the teacher role, including face features, body features, the confidence levels corresponding to the face features and body features respectively, and teacher information in the class schedule. The face features can be the facial feature information of the teacher. The body features can be the physical features, body shape, posture, clothing worn, etc. of the teacher. The confidence levels corresponding to the face features and body features respectively refer to the recognition confidence levels of each feature, which are used to measure the accuracy of the features. The teacher information in the class schedule can include the identity information of the teacher, such as name, teacher ID, job title, etc.

[0043] Among them, the calculation method of the confidence levels corresponding to the face features and body features respectively can be: single class period confidence level = the number of occurrences of a certain feature (face feature or body feature) in this class period / total number of valid frames; the valid frame refers to a frame in each frame image that contains at least one face feature or body feature information; the total number of valid frames is the sum of the quantities of all valid frames.

[0044] The teacher feature group is used to identify and confirm the identity of the teacher. In the classroom, by comparing the features extracted in the current classroom with the teacher feature group in the database, it is determined whether the feature belongs to a teacher that has been stored. The teacher feature group plays a very crucial role in teacher role recognition. By initializing the teacher feature group, accurate identity recognition can be performed based on the features of the teacher in the classroom (such as face features and body features), and the feature library can be continuously optimized and updated to cope with various possible changes in the classroom environment.

[0045] Specifically, see Figure 3 , the initialization of the teacher feature group includes:

[0046] S111. Trigger the start of the teaching analysis device based on the class schedule time, obtain the teacher data of the classroom through the started teaching analysis device, and obtain the target teacher features according to the teacher data. The target teacher features are the face features and / or body features with the highest frequency of occurrence in the teacher data.

[0047] At the beginning of each class, the system that executes the teacher role recognition method can automatically start the teaching analysis device according to the class schedule time and begin to collect the feature data in the classroom, so as to ensure that the system can obtain the teacher data of the current classroom in real time and then obtain the target teacher features. For example, after the teaching analysis device (which may include the video acquisition device, computing device, and storage device in the above application scenario) is started, algorithms such as face recognition and / or person re-identification are used to capture in real time the feature that appears most frequently in the classroom (for example, the face feature or body feature of the teacher). The feature that appears most frequently is usually the feature with a relatively high correlation with the teacher's identity. Among them, the teaching analysis device starts real-time video acquisition to obtain the teacher data in the classroom scene. Face data and body data are obtained according to the teacher data; a face feature vector is obtained according to the face data, and a body feature vector is obtained according to the body data. The body feature vector includes the body outline feature, clothing color feature, and bone key point feature; the number of occurrences of each face feature vector is counted, and similar face features are clustered to obtain the total frequency of each face feature cluster; the number of occurrences of each body feature vector is counted, and similar body features are clustered to obtain the total frequency of each body feature cluster; according to the total frequency of each face feature cluster and the total frequency of each body feature cluster, the face feature vector and / or body feature vector that appears most frequently within the corresponding duration of the class is selected as the target teacher feature. Among them, the teacher data can be considered as the data collected by the teaching analysis device for the podium area. The camera, depth sensor, etc. of the teaching analysis device are mainly aimed at the podium area for data collection to ensure that the facial and body features of the teacher can be captured. Through face detection technology, all face data is extracted from the video frames of the podium area, and the face feature vector is calculated. Through human body detection technology, information such as the body outline, clothing color, and bone key points is extracted from the video to generate a body feature vector.

[0048] S112. If the target teacher feature is not included in the preset database, then the target teacher feature, the corresponding feature confidence, and the teacher information in the class schedule are taken together as the first group of teacher features, and the first group of teacher features is stored in the database.

[0049] Among them, by comparing the features extracted in the current class with the features in the existing feature library, it is confirmed whether there are the features of this teacher. If not, the currently obtained features (face features and / or body features) and the corresponding confidence (that is, the accuracy of the features) and the teacher information in the class schedule are stored in the database together to form the first group of teacher features. Thus, it is ensured that when new teacher features are encountered for the first time, these feature information can be stored in the database in a timely manner. If the feature is not in the database, the system will regard it as a new teacher feature and save it.

[0050] S113. Trigger the teaching analysis device to start again based on the class schedule time, and obtain the teacher data of the class through the started teaching analysis device, and obtain the above-mentioned target teacher features again according to the teacher data.

[0051] Similar to step S111, the system obtains the face feature or body feature with the highest frequency of occurrence in the current class session again. By triggering the start of the teaching analysis device again, the system can continuously collect data during the teacher's class to ensure that the number of the obtained teacher feature groups meets the requirements.

[0052] S114. Compare the target teacher features obtained again with the first group of teacher features. If the similarity is less than or equal to the preset threshold, then take the target teacher features obtained again, as well as the corresponding feature confidence and the teacher information in the class schedule, together as the second group of teacher features, and store the second group of teacher features in the database.

[0053] Among them, when performing a group of feature comparisons, the face features can be compared first, and then the body features.

[0054] Among them, the preset threshold is used to determine whether the newly extracted features are close enough to the stored features, so as to determine whether they are regarded as the features of the same teacher. The setting of the preset threshold directly affects the recognition accuracy and misrecognition rate of the system. The preset threshold can be set according to the stability and distinctiveness of the features. For example, the features of a teacher (such as facial and body features) usually remain stable in different classroom environments. The system should ensure that the features extracted in multiple classes of the same teacher have sufficient stability. Even if there are some small changes (such as different perspectives, etc.), these features should still be able to be recognized as the features of the same teacher. If the features of the teacher itself have high distinctiveness (such as, the facial features are particularly obvious or the body shape features are significant), then the preset threshold can be relatively high because similar features are not easily confused; but if the teacher's appearance is relatively ordinary and the appearance of other people in the class is similar to that of the teacher, then the preset threshold needs to be set lower to avoid misrecognition. The preset threshold can be a fixed threshold or a dynamic threshold. During the continuous recognition process, the system adjusts the threshold according to the actual situation. For example, if there are more misrecognitions, the threshold can be dynamically reduced; on the contrary, the threshold can be increased to reduce missed recognitions.

[0055] Among them, the newly extracted features are compared with the existing first group of teacher features in the database through similarity comparison. If the similarity is less than or equal to the preset threshold (that is, the new features are quite different from the existing features), it is considered that the features may belong to another teacher role or different features of the teacher (for example, the teacher wears different clothes at different times, the facial angle changes, etc.). The system will store the new features and related information together as the second group of teacher features, thereby improving the stability and accuracy of verifying teacher features.

[0056] Among them, if the above similarity is greater than a preset threshold, the method further includes: determining the object corresponding to the most frequently occurring face feature and / or human body feature obtained again as the teacher role, and at the same time increasing the confidence level corresponding to the first set of teacher features. By comparing the newly obtained features with the existing features, the teacher identity can be confirmed more accurately, the possibility of misidentification can be reduced, and increasing the feature confidence level enables dynamic adaptation to changes in the teacher identity, enhancing the reliability and applicability of the feature group.

[0057] S115. At different time points when the teacher is having class, continuously start the teaching analysis device to obtain the target teacher features in the classroom, and compare the obtained target teacher features with the teacher features in the currently stored teacher feature group. If the similarity is less than or equal to the preset threshold, store the newly obtained target teacher features, the corresponding feature confidence level, and the teacher information in the class schedule as a new teacher feature group in the database, and repeat the above process until the number of teacher feature groups reaches the preset number.

[0058] At different time points when the teacher is having class, the system will continuously start the teaching analysis device and obtain new features. Each time the newly obtained features are compared with the existing teacher feature group. If the similarity is low (i.e., there are significant differences between the new features and the existing features), the system adds them as new features and updates the teacher feature group. This process will continue until the teacher feature group reaches the set number (such as three groups of teacher features). During this process, the feature information will be continuously optimized and updated to ensure higher recognition accuracy.

[0059] The above steps S111 to S115 establish a dynamically updated teacher feature database to ensure the continuous accumulation and accurate matching of the teacher's identity information in different classrooms. Through this dynamic update mechanism, more accurate teacher features can be accumulated during the teacher's multiple teaching sessions, further improving the recognition accuracy and ensuring the dynamic update of the feature database, thus achieving accurate teacher role recognition. In addition, this automated and dynamic update mechanism enables the teacher not to actively provide data, and can automatically extract and optimize features from the classroom, simplifying the teacher management process, while enhancing the reliability and security of the system.

[0060] It should be noted that the above step S11 is a process of initializing the teacher feature group, which is not a step that needs to be executed every time in the teacher role recognition method. It can be executed when the system runs for the first time or when the teacher feature group needs to be updated. At the beginning of each class, the system will still start the teaching analysis device, but it does not need to re-initialize the teacher feature group. Instead, it can directly obtain the teacher features in the classroom and compare them with the teacher feature group in the database. If the matching degree between the teacher features detected in the current class and the existing teacher feature group is low, the system can add new teacher features (such as a new teacher if the current teacher is not in the database) or replace the low-confidence features (such as when the existing teacher features change, like the teacher changes clothes). This update mechanism can make the teacher feature group more stable and accurate with the accumulation of classroom data, thus improving the accuracy of teacher role recognition.

[0061] S12. Obtain the teacher data in the current class and identify the first teacher features based on this teacher data.

[0062] Among them, the teacher data is extracted from the current class, and the first teacher features are identified based on this teacher data, that is, the preliminary feature information of the teacher identified in the class. This preliminary feature information can include face features or body features, etc. The obtained teacher features are the basis for subsequent comparison with the existing features in the feature group, and this process determines whether the teacher identity can be successfully matched.

[0063] S13. Compare the first teacher features with each group of teacher features in the teacher feature group one by one, and calculate the similarity between the first teacher features and each group of teacher features in the teacher feature group.

[0064] Among them, the first teacher features are compared with each group of teacher features stored in the database one by one, and the similarity between each group of features is evaluated using a similarity calculation method (such as Euclidean distance, cosine similarity, etc.).

[0065] S14. If the similarity between the first teacher features and the teacher features in any group is greater than the preset threshold, determine that the object corresponding to the first teacher features is the teacher role.

[0066] Among them, if the similarity exceeds the threshold, it indicates that the identified features conform to the teacher identity, and it can be confirmed that the features represent the teacher role, and these features can continue to be used to confirm the subsequent teacher identity.

[0067] Among them, the face feature and / or body feature with the highest occurrence frequency is obtained from the first teacher feature; the face feature and / or body feature with the highest occurrence frequency is compared with the teacher features in the teacher feature group group by group for similarity. If the similarity of the teacher feature in any group is greater than the preset threshold, it is determined that the object corresponding to the face feature and / or body feature with the highest occurrence frequency in the first teacher feature is the teacher role.

[0068] Among them, when the comparison result shows that the first teacher feature of the current class is highly similar to a certain teacher feature in the database (that is, the similarity is greater than the preset threshold), not only is it confirmed that this person is the teacher role, but there can also be more information associated with the teacher role. Since the teacher feature group usually contains other information related to the teacher, such as the teacher's name, work number, teaching subject, etc., and these information are stored by associating the teacher information in the class schedule with the feature data, the information associated with the teacher feature group also includes information such as the teaching time and course arrangement. Therefore, after identifying the teacher role, the teacher identity information corresponding to the teacher role can also be obtained.

[0069] Among them, if the similarity between the first teacher feature and the teacher feature in any group is greater than the preset threshold, the confidence level corresponding to the teacher feature group with a similarity greater than the preset threshold to the first teacher feature is increased.

[0070] S15. If the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, enter the dynamic replacement process.

[0071] If the similarity of all feature groups does not exceed the preset threshold, the system will trigger the dynamic replacement process. The dynamic replacement process will re-evaluate the feature with the highest occurrence frequency extracted in the current class and update the teacher feature group with the lowest confidence level in the database with it as the new teacher feature information. This step ensures that when existing features cannot be matched, it can be automatically adjusted according to the data in the class and gradually update the teacher feature library, which can effectively avoid the situation of missed recognition.

[0072] Among them, the dynamic replacement process includes: using the feature with the highest occurrence frequency in the teacher features identified in the current class as the teacher feature information, and replacing the group of teacher features with the lowest confidence level in the database with the teacher feature information to obtain the updated teacher feature group.

[0073] Among them, the process of determining whether the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to a preset threshold may include: obtaining the face feature and / or body feature with the highest occurrence frequency, as well as the face feature and / or body feature with the second and third highest occurrence frequencies, from the first teacher feature; if the similarity between the face feature and / or body feature with the highest occurrence frequency and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, then comparing the similarity between the face feature and / or body feature with the second and third highest occurrence frequencies and each group of teacher features in the teacher feature group respectively; if the results of the similarity comparison are all less than or equal to the preset threshold, it is determined that the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold.

[0074] In the above teacher role recognition process, through hierarchical matching and multi-level feature comparison, the recognition accuracy of the system in a complex environment is improved. First, the face feature and / or body feature with the highest occurrence frequency is extracted from the first teacher feature identified in the current classroom, because the feature with a higher frequency is usually the most stable and representative, and may best reflect the identity of the teacher. Further extracting the face feature and / or body feature with the second and third highest occurrence frequencies means that when recognizing teacher features, not only a single feature is relied on, but multiple features are considered. This multi-dimensional feature extraction can increase the robustness of the system and avoid misrecognition due to the absence or instability of a certain feature.

[0075] The similarity between the features with the highest, second highest, and third highest occurrence frequencies extracted is compared with each group of teacher features in the database respectively. Through such one-by-one comparison, the system can more comprehensively evaluate the matching situation between the current feature and the teacher feature group. Moreover, even if a feature fails to be recognized due to certain reasons (such as angle problems, occlusion, etc.), the second highest or third highest feature may still provide effective clues to ensure the recognition accuracy.

[0076] In some embodiments, continue to refer to Figure 2 , the method further includes:

[0077] S16. In the next class, obtain the second teacher feature identified in the next class, and compare the similarity between the second teacher feature and each group of the updated teacher feature group one by one. If the similarity is greater than the preset threshold, it is determined that the object corresponding to the second teacher feature is a teacher role; if the similarity is less than or equal to the preset threshold, execute the above dynamic replacement process.

[0078] Among them, the updated teacher feature group will continue to be used to calculate the similarity with the new second teacher feature. If the comparison similarity exceeds the preset threshold, the system confirms that the object corresponding to this feature is a teacher role; if the similarity does not meet the preset threshold, the dynamic replacement process will be executed to continue updating the feature library. This step ensures that the teacher feature group can be continuously optimized through continuous analysis across class sessions. When the system obtains more class session data, it can gradually optimize the teacher feature library, improve the stability and accuracy of the system. Through repeated comparison and update, the recognition process will become more accurate.

[0079] It should be noted that after the above step S14 determines that the object corresponding to the first teacher feature is a teacher role, the above step S16 can be executed to further verify the teacher role in the next class session. After the above step S15 determines that the object corresponding to the first teacher feature is not a teacher role and enters the dynamic replacement process, the above step S16 can also be executed to continue adjusting and updating the teacher feature group in subsequent class sessions. In addition, step S16 can be repeatedly executed. Specifically, step S16 describes that in the next class session, the system will obtain new teacher features and compare them with the currently stored teacher feature group. This process is a continuous and dynamic recognition process, so it can be repeatedly executed in subsequent class sessions to continuously update and improve the teacher feature group and ensure that the system can accurately identify the teacher role based on the latest data.

[0080] Among them, refer to Figure 4 , compare the second teacher feature with the updated teacher feature group group by group. If the similarity is greater than the preset threshold, determining that the object corresponding to the second teacher feature is a teacher role includes:

[0081] S161. Obtain the face feature and / or body feature with the highest occurrence frequency from the second teacher feature.

[0082] S162. Compare the face feature and / or body feature with the highest occurrence frequency with the updated teacher feature group group by group. If the similarity is greater than the preset threshold, determine that the object corresponding to the face feature and / or body feature with the highest occurrence frequency in the second teacher feature is a teacher role.

[0083] S163. Increase the corresponding confidence level of the teacher feature group with a similarity greater than the preset threshold to the face feature and / or body feature with the highest occurrence frequency.

[0084] Among them, obtain the current confidence corresponding to the teacher feature group whose similarity to the most frequently occurring face feature and / or human body feature is greater than a preset threshold; sum the current confidence corresponding to the teacher feature group whose similarity is greater than the preset threshold with the confidence corresponding to the second teacher feature to obtain a new confidence; use the calculated new confidence as the confidence of the teacher feature group whose similarity is greater than the preset threshold and update it. The update of this confidence means that the system gradually increases its trust in a specific teacher feature group, reflecting the improvement in the accuracy of the system's recognition of the teacher role. In particular, when there is more and more new teacher feature data, the accumulation of confidence can effectively help the system identify the teacher role.

[0085] In this embodiment, as each class progresses, the system not only continuously identifies the teacher role, but also can more accurately identify and distinguish the teacher by updating the confidence of the teacher features. The method of this embodiment enhances the flexibility, stability, and accuracy of teacher role recognition, while reducing the possibility of misrecognition.

[0086] The following illustrates the teacher role recognition method of the embodiments of the present application by way of examples.

[0087] For example: There is a classroom teaching scenario where the system is performing teacher role recognition. The scenario settings include Teacher Li for a math class. The system uses a camera to capture the classroom video and extracts the teacher's features through visual algorithms such as face recognition and person re-identification. First, initialize the teacher feature group. When the first math class of Teacher Li starts and the schedule time arrives, the system is automatically started by the teaching analysis device. The system captures the classroom scene through the camera, extracts the teacher data in the classroom, and identifies the most frequently occurring face feature and human body feature based on the teacher data. The system finds through analysis that Teacher Li's face feature (such as facial features) occurs most frequently. If these features have not been stored in the database, the system stores these most frequently occurring features together with Teacher Li's schedule information as the first group of teacher features and stores them in the database. The system starts the teaching analysis device again in a subsequent class (such as Teacher Li's second math class) to obtain the teacher features in the classroom. In the second class, Teacher Li's facial features and body shape features appear again. The system compares these features with the first group of teacher features and finds that the similarity is higher than the preset threshold, so the first group of teacher features remains unchanged; if the system finds that the similarity is lower than the preset threshold, the feature group will be updated.

[0088] Next, teacher role recognition and dynamic replacement are performed. In a certain class, the system continues to extract the facial features and body features of the target object as the first teacher features. The system compares the similarity between the first teacher features and the first set of teacher features stored in the teacher feature group. Suppose the similarity between the facial features in the first teacher features and the first set of teacher features is greater than the preset threshold. The system confirms that the feature matches the identity of Teacher Li, that is, the object corresponding to the first teacher features is a teacher role and is Teacher Li. Suppose the similarity between the first teacher features and the first set of teacher features is less than or equal to the preset threshold, then the dynamic replacement process is entered. Specifically, the system replaces the feature with the lowest confidence in the teacher feature group with the feature with the highest frequency in the current class (which may be a newly emerged facial feature or body feature). The updated teacher feature group will be stored in the database. Next, in a certain class, the system starts again and extracts the second teacher features. Through analysis, the system finds that the new teacher features (such as body type features) do not exactly match the first set of teacher features, but have a high similarity with the features of Teacher Li in the updated teacher feature group. Then, the object corresponding to the second teacher features can be determined as a teacher role and is Teacher Li. As the teacher's teaching progresses, the system continuously optimizes the teacher feature group through the feature data of each class, ensuring the accuracy of teacher role recognition. The feature comparison and confidence update of each class ensure that the system can effectively identify Teacher Li and gradually reduce the probability of misidentification during the teaching process. Moreover, by gradually updating the teacher feature group, the system can more accurately identify the teacher role in subsequent classes, and the recognition effect is continuously improved as the data accumulates.

[0089] This embodiment demonstrates how to ensure the accurate recognition of the teacher role through data comparison, feature extraction, and dynamic update of the teacher feature group in multiple classes. In each class, the system identifies based on the teacher's facial, body type, and other features, and continuously optimizes the teacher feature group through similarity comparison and confidence update, so as to ensure the correct identification of the teacher's identity.

[0090] See Figure 5 , Figure 5 which is a schematic structural diagram of a teacher role recognition device provided by an embodiment of the present application. The device 20 includes:

[0091] A teacher feature group acquisition module 21, configured to initialize the teacher feature group, where the teacher feature group includes at least one set of teacher features and is stored in a preset database;

[0092] A first teacher feature acquisition module 22, configured to acquire teacher data in the current class and identify the first teacher features based on the teacher data;

[0093] The first comparison processing module 23 is configured to compare the first teacher feature with each group of teacher features in the teacher feature group one by one, and calculate the similarity between the first teacher feature and each group of teacher features in the teacher feature group;

[0094] The teacher role determination module 24 is configured to determine that the object corresponding to the first teacher feature is a teacher role if the similarity between the first teacher feature and the teacher feature in any group is greater than a preset threshold;

[0095] The dynamic replacement module 25 is configured to enter a dynamic replacement process if the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold; wherein, the dynamic replacement process includes: using the feature with the highest occurrence frequency among the teacher features identified in the current class as the teacher feature information, and replacing the group of teacher features with the lowest confidence in the database with the teacher feature information to obtain an updated teacher feature group;

[0096] The second comparison processing module 26 is configured to, in the next class, obtain the second teacher feature identified in the next class, compare the second teacher feature with each group in the updated teacher feature group for similarity. If the similarity is greater than the preset threshold, it is determined that the object corresponding to the second teacher feature is a teacher role; if the similarity is less than or equal to the preset threshold, the dynamic replacement process is executed.

[0097] Among them, the above-mentioned teacher feature group acquisition module 21 is used to execute the process of initializing the teacher feature group. This process is not a step that needs to be executed every time in the teacher role recognition method, and it can be executed when the system runs for the first time or when the teacher feature group needs to be updated. In addition, the function of the second comparison processing module 26 is to perform teacher feature comparison in the next class, which is used to continuously verify and optimize the accuracy of teacher recognition. The second comparison processing module 26 is used as an example, and its function is similar to that of the teacher role determination module 24 and the dynamic replacement module 25.

[0098] Among them, the above-mentioned teacher role recognition device 20 can be a software module. The software module includes several instructions, which are stored in the memory. The processor can access this memory and call the instructions for execution to complete the teacher role recognition method described in the above various embodiments.

[0099] In some embodiments, the above-mentioned teacher role recognition device 20 can also be built by hardware devices. For example, the teacher role recognition device 20 can be built by one or more than two chips, and each chip can work in coordination with each other to complete the above-mentioned teacher role recognition method described in each of the above embodiments. For another example, the above-mentioned teacher role recognition device 20 can also be built by various logic devices, such as being built by a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination of these components.

[0100] It should be noted that the above-mentioned teacher role recognition device 20 can execute the above-mentioned teacher role recognition method applied to an electronic device provided in the embodiments of the present application, and has corresponding functional modules and beneficial effects for executing the method. For technical details not described in detail in the embodiments of the teacher role recognition device 20, reference can be made to the above-mentioned teacher role recognition method applied to an electronic device provided in the embodiments of the present application.

[0101] See Figure 6 , Figure 6 is a schematic structural diagram of an electronic device provided in an embodiment of the present application. The electronic device 30 includes one or more processors 31 and a memory 32. The memory 32 is connected to one or more processors 31, for example, connected to the processor 31 through a bus.

[0102] The processor 31 is configured to support the electronic device 30 to execute the corresponding functions in the method in the above-mentioned method embodiments. The processor 31 can be a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof. The above-mentioned hardware chip can be an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above-mentioned PLD can be a complex programmable logic device (CPLD), a field programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0103] The memory 32 is used to store program codes and the like. The memory 32 may include volatile memory (VM), such as random access memory (RAM); the memory may also include non-volatile memory (NVM), such as read-only memory (ROM), flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 32 may also include a combination of the above types of memories.

[0104] The memory 32 can be used to store non-volatile software programs, non-volatile computer-executable programs and modules, such as the program instructions / modules corresponding to the teacher role recognition method in the embodiments of the present application. The processor 31 executes various functional applications and data processing of the teacher role recognition method and the teacher role recognition device by running the non-volatile software programs, instructions and modules stored in the memory 32, that is, realizes the functions of the various modules or units of the teacher role recognition method and the teacher role recognition device provided in the above method embodiments.

[0105] The memory 32 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function. The data storage area can store data created according to the use of the teacher role recognition device. In some embodiments, the memory 32 may include a memory remotely set relative to the processor, and these remote memories can be connected to the teacher role recognition device through a network. Examples of the above network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network and their combination.

[0106] The one or more modules are stored in the memory 32, and when executed by the one or more processors 31, execute the teacher role recognition method in any of the above method embodiments. For example, execute the method steps described in the above method embodiments to realize the functions of the modules described in the above device embodiments.

[0107] The electronic device in the embodiments of the present application may specifically be an ultra-mobile personal computer device, a server or a server cluster, etc.

[0108] The embodiments of the present application provide a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores computer-executable instructions, and these computer-executable instructions are executed by one or more processors, for example Figure 6One of the processors 31 can enable the above-mentioned one or more processors to execute the teacher role recognition method in any of the above method embodiments. For example, execute the Figures 2 to 4 method steps described above to implement Figure 5 the functions of the modules in

[0109] An embodiment of the present application provides a computer program product. The computer program product includes a computer program stored on a non-volatile computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by the electronic device, the electronic device can execute the teacher role recognition method in any of the above method embodiments. For example, execute the Figures 2 to 4 method steps described above to implement Figure 5 the functions of the modules in

[0110] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0111] The above-disclosed are only the preferred embodiments of the present application. Of course, the scope of the rights of the present application cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present application still fall within the scope covered by the present application.

Claims

1. A method for identifying teacher roles, characterized in that, Including: Obtain teacher data in the current class, and identify the first teacher feature according to the teacher data; Compare the first teacher feature with each group of teacher features in the teacher feature group one by one, and calculate the similarity between the first teacher feature and each group of teacher features in the teacher feature group; the teacher feature group includes at least one group of teacher features and is stored in a preset database; If the similarity between the first teacher feature and the teacher feature in any group is greater than the preset threshold, determine that the object corresponding to the first teacher feature is a teacher role; If the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, enter the dynamic replacement process; wherein, the dynamic replacement process includes: using the feature with the highest occurrence frequency in the teacher features identified in the current class as the teacher feature information, and replacing the group of teacher features with the lowest confidence in the database with the teacher feature information to obtain an updated teacher feature group.

2. The method according to claim 1, wherein The method further includes: initializing the teacher feature group; The initializing the teacher feature group includes: Trigger the teaching analysis device to start based on the class schedule time, obtain the teacher data of the class through the started teaching analysis device, and obtain the target teacher feature according to the teacher data, where the target teacher feature is the face feature and / or body feature with the highest occurrence frequency in the teacher data; If the target teacher feature is not included in the preset database, use the target teacher feature, the corresponding feature confidence, and the teacher information in the class schedule as the first group of teacher features, and store the first group of teacher features in the database; Trigger the teaching analysis device to start again based on the class schedule time, and obtain the teacher data of the class through the started teaching analysis device, and obtain the target teacher feature again according to the teacher data; Compare the target teacher feature obtained again with the first group of teacher features. If the similarity is less than or equal to the preset threshold, use the target teacher feature obtained again, the corresponding feature confidence, and the teacher information in the class schedule as the second group of teacher features, and store the second group of teacher features in the database; At different time points when the teacher is in class, continuously start the teaching analysis device, obtain the target teacher feature in the class, and compare the obtained target teacher feature with the teacher feature in the currently stored teacher feature group. If the similarity is less than or equal to the preset threshold, use the newly obtained target teacher feature, the corresponding feature confidence, and the teacher information in the class schedule as a new teacher feature group and store it in the database, and repeat the above process until the number of teacher feature groups reaches the preset number.

3. The method according to claim 2, wherein Compare the target teacher feature obtained again with the first group of teacher features. If the similarity is greater than the preset threshold, the method further includes: Determine the object corresponding to the target teacher feature obtained again as the teacher role, and at the same time increase the confidence corresponding to the first group of teacher features.

4. The method according to claim 2, wherein Obtaining the target teacher features according to the teacher data, where the target teacher features are the face features and / or body features with the highest frequency of occurrence in the teacher data, including: Obtaining face data and body data according to the teacher data; Obtaining face feature vectors according to the face data, and obtaining body feature vectors according to the body data; the body feature vectors include body outline features, clothing color features, and bone key point features; Counting the occurrence times of each face feature vector, and clustering similar face features to obtain the total frequency of each face feature cluster; Counting the occurrence times of each body feature vector, and clustering similar body features to obtain the total frequency of each body feature cluster; According to the total frequency of each face feature cluster and the total frequency of each body feature cluster, selecting the face feature vector and / or body feature vector with the most occurrence times within the corresponding duration of the class as the target teacher features.

5. The method according to claim 1, characterized in that, The similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to a preset threshold, including: Obtaining the face features and / or body features with the highest frequency of occurrence from the first teacher feature, as well as the face features and / or body features with the second and third highest frequencies of occurrence; If the similarity between the face features and / or body features with the highest frequency of occurrence and each group of teacher features in the teacher feature group is less than or equal to the preset threshold, then compare the similarity between the face features and / or body features with the second and third highest frequencies of occurrence and each group of teacher features in the teacher feature group respectively; If the results of the similarity comparison are all less than or equal to the preset threshold, it is determined that the similarity between the first teacher feature and each group of teacher features in the teacher feature group is less than or equal to the preset threshold.

6. The method according to any one of claims 1 to 5, characterized in that, If the similarity between the first teacher feature and the teacher feature in any one group is greater than the preset threshold, determining that the object corresponding to the first teacher feature is a teacher role, including: Obtaining the face features and / or body features with the highest frequency of occurrence from the first teacher feature; Comparing the similarity between the face features and / or body features with the highest frequency of occurrence and the teacher features in each group of the teacher feature group one by one. If the similarity between the teacher feature in any one group is greater than the preset threshold, it is determined that the object corresponding to the face features and / or body features with the highest frequency of occurrence in the first teacher feature is a teacher role.

7. The method according to claim 6, wherein If the similarity between the first teacher feature and the teacher feature in any one group is greater than the preset threshold, the method further includes: Increasing the corresponding confidence level of the teacher feature group with a similarity greater than the preset threshold to the first teacher feature.

8. The method according to claim 7, wherein Increasing the corresponding confidence level of the teacher feature group with a similarity greater than the preset threshold to the first teacher feature, including: Obtaining the current confidence level corresponding to the teacher feature group with a similarity greater than the preset threshold to the face features and / or body features with the highest frequency of occurrence; Sum the current confidence corresponding to the group of teacher features with a similarity greater than the preset threshold and the confidence corresponding to the first teacher feature to obtain a new confidence; Use the calculated new confidence as the confidence of the group of teacher features with a similarity greater than the preset threshold and update it.

9. An electronic device, characterized in that, Including: A memory and a processor, the memory is connected to the processor, the processor is configured to execute one or more computer programs stored in the memory, and when the processor executes the one or more computer programs, the electronic device implements the method according to any one of claims 1-8.

10. A non-volatile computer-readable storage medium, characterized in that, The non-volatile computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by an electronic device, the electronic device executes the method according to any one of claims 1-8.

Images