Training room management platform and method based on artificial intelligence

Abstract

The present application relates to the field of training room management, and specifically discloses an artificial intelligence-based training room management platform and method, which first obtains the training room monitoring image collected by a camera, the course information collected by a database, and the student identity information collected by an RFID reader installed in the training room, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a classifier to determine whether to issue a suspicious person warning alarm, so as to help managers better understand the operation of the training room, thereby strengthening the safety prevention capabilities of the training room and improving the overall safety level of the training room.

Description

Technical Field

[0001] The present application relates to the field of training room management, and more specifically, to an artificial intelligence-based training room management platform and method. Background Art

[0002] The goal of today's higher education is not only to impart theoretical knowledge, but more importantly to cultivate compound talents with hands-on and brain-powered abilities. As the main place for colleges and universities to cultivate students' practical skills, the management of the training room is crucial. The training room is a place provided by schools or enterprises for students to practice operations and skills training. It is usually equipped with various professional equipment and tools to provide opportunities for practical operations, experiments and skills training, helping students gain practical experience and skills in specific fields.

[0003] However, the existing training room safety management system has some shortcomings. Most training room safety management systems rely mainly on surveillance cameras for management, which requires manual viewing of surveillance images. This method is time-consuming and labor-intensive, requiring personnel to watch the surveillance images for a long time, and is prone to safety management loopholes due to blind spots in monitoring, affecting the safety management of the training room. In addition, this safety management method has a low level of intelligence and is overly dependent on manual operation.

[0004] Therefore, an artificial intelligence-based training room management platform and method are desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a training room management platform and method based on artificial intelligence, which first obtains the training room monitoring image collected by the camera, the course information collected by the database and the student identity information collected by the RFID reader installed in the training room, and then uses deep learning technology to perform feature extraction and association analysis on the three, and finally uses a classifier to determine whether to issue a suspicious person warning alarm, so as to help managers better understand the operation of the training room, thereby strengthening the safety prevention capability of the training room and improving the overall safety level of the training room.

[0006] According to one aspect of the present application, there is provided an artificial intelligence-based training room management platform, which includes:

[0007] The training room management data acquisition module is used to obtain the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room;

[0008] A training room management data extraction module is used to extract optimized training room monitoring significant feature vectors and training room student identity fusion feature vectors from the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room;

[0009] The suspicious person early warning judgment module is used to judge whether to issue a suspicious person early warning alarm based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector.

[0010] According to another aspect of the present application, there is provided a training room management method based on artificial intelligence, which includes:

[0011] Obtain the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room;

[0012] Extracting optimized training room monitoring significant feature vectors and training room student identity fusion feature vectors from the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader / writer installed in the training room;

[0013] Based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector, it is determined whether to issue a suspicious person early warning alarm.

[0014] Compared with the prior art, the present application provides an artificial intelligence-based training room management platform and method, which first obtains the training room monitoring image collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room, and then uses deep learning technology to perform feature extraction and association analysis on the three, and finally uses a classifier to determine whether to issue a suspicious person warning alarm, so as to help managers better understand the operation of the training room, thereby strengthening the safety prevention capabilities of the training room and improving the overall safety level of the training room. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of an artificial intelligence-based training room management platform according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a training room management data extraction module in an artificial intelligence-based training room management platform according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a training room monitoring image feature extraction unit in an artificial intelligence-based training room management platform according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a suspicious person early warning judgment module in an artificial intelligence-based training room management platform according to an embodiment of the present application.

[0020] Figure 5 The present invention is a flowchart of an artificial intelligence-based training room management method according to an embodiment of the present application.

[0021] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0023] Figure 1 FIG. 1 is a block diagram of an artificial intelligence-based training room management platform according to an embodiment of the present application. Figure 1 As shown, according to the embodiment of the present application, the artificial intelligence-based training room management platform 100 includes: a training room management data acquisition module 110, which is used to obtain the training room monitoring image collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; a training room management data extraction module 120, which is used to extract the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector from the training room monitoring image collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; a suspicious person early warning judgment module 130, which is used to judge whether to issue a suspicious person early warning alarm based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector.

[0024] In the above-mentioned training room management platform 100 based on artificial intelligence, the training room management data acquisition module 110 is used to obtain the training room monitoring image collected by the camera, the course information collected by the database and the student identity information collected by the RFID reader installed in the training room. It should be understood that when talking about the goals of higher education today, it is no longer just emphasizing the imparting of theoretical knowledge, but more focusing on cultivating students with hands-on and brain-powered comprehensive talents. As the main place for schools to provide students with practical skills training, the management of the training room is particularly important. The training room is usually equipped with various professional equipment and tools, providing practical operation, experimental and skills training opportunities, aiming to help students accumulate practical experience and skills in specific fields. However, there are some shortcomings in the existing training room safety management system. At present, most training room safety management systems mainly rely on surveillance cameras for management, and manual continuous viewing of surveillance images is required. This method is not only time-consuming and laborious, but also requires personnel to stare at the surveillance images for a long time. It is also easy to have security management loopholes due to blind spots in monitoring, affecting the safety control of the training room. In addition, this safety management model has a low level of intelligence, relies too much on manual operation, and lacks an efficient automated management mechanism. Therefore, in the technical solution of the present application, by obtaining the monitoring images of the training room captured by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room, and combining it with deep learning technology, it is judged whether to issue a suspicious person warning alarm, thereby performing security management and control of the training room.

[0025] Specifically, the monitoring images of the training room collected by the camera provide real-time visual information, which can help monitor the activities and situations in the training room. By analyzing the monitoring images, real-time monitoring, anomaly detection, personnel identification and other functions can be realized, thereby ensuring the safety and order of the training room. The course information collected by the database contains important information such as the arrangement, content and time of various courses in the training room. This information is crucial for students' learning and teachers' teaching management. By obtaining course information, managers can make corresponding adjustments according to the course schedule to improve teaching efficiency and quality. The student identity information collected by the RFID reader installed in the training room can help realize the automatic identification and management of student identities. The identity information of each student can be uniquely identified and recorded, thereby realizing functions such as student attendance, entry and exit management, and behavior analysis. This not only improves the convenience of student management, but also helps to ensure the safety and order of the training room. In this way, comprehensive monitoring and management of the training room management system can be achieved. Through the application of artificial intelligence technology, this information can be intelligently analyzed and processed to realize the intelligent and automated management of the training room, improve management efficiency, strengthen safety precautions, and provide students and teachers with a better learning and teaching environment.

[0026] In the above-mentioned training room management platform 100 based on artificial intelligence, the training room management data extraction module 120 is used to extract the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector from the training room monitoring image collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room. It should be understood that the purpose of extracting the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector from the training room monitoring image collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room is to combine multi-source data and realize more comprehensive and accurate training room management and monitoring through artificial intelligence technology. In this way, the training room management system can more comprehensively understand the operating status of the training room and the activity trajectory of the students, realize intelligent management and monitoring of the training room, and provide decision support, thereby improving the safety, efficiency and management level of the training room. This method of comprehensive utilization of multi-source data helps to realize the intelligent and personalized management of the training room, and provide better management tools and services for education and training institutions.

[0027] Figure 2 FIG. 1 is a block diagram of a training room management data extraction module in an artificial intelligence-based training room management platform according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the training room management data extraction module 120 includes: a training room monitoring image feature extraction unit 121, which is used to perform feature extraction on the training room monitoring image collected by the camera to obtain the optimized training room monitoring significant feature vector; a training room course information feature extraction unit 122, which is used to perform feature extraction on the training room course information collected by the database to obtain a semantic global feature vector of the training room course information; a student identity information feature extraction unit 123, which is used to perform feature extraction on the student identity information collected by the RFID reader installed in the training room to obtain a student identity associated feature vector; a student identity feature fusion unit 124, which is used to fuse the training room course information semantic global feature vector and the student identity associated feature vector to obtain the training room student identity fusion feature vector.

[0028] It should be understood that training room monitoring images usually contain a lot of information, including students, teachers, equipment and other elements. Through feature extraction, these complex image information can be converted into more representative and easy-to-process feature vectors. Among them, optimizing the extraction of significant feature vectors for training room monitoring can help identify key features in training room monitoring images, such as personnel behavior, abnormal conditions, object locations, etc., thereby helping to improve the safety and management efficiency of training rooms.

[0029] Furthermore, by extracting features from course information, various attributes of the course, related information, and student participation can be converted into numerical feature vectors. Such feature vectors not only contain basic information such as course name, time, and location, but also include deeper semantic information such as course content, objectives, and difficulty. This global feature vector extraction can be achieved through natural language processing technology, text mining, and machine learning methods, making course information easier to understand, compare, and use.

[0030] Furthermore, the student identity information collected by the RFID reader contains the student's unique identification, such as the student ID or card number. In this way, these identity information can be converted into feature vectors with semantic meaning, so that the student identity can be more effectively identified and associated. Here, the extraction of student identity-related feature vectors helps to track and analyze the student's behavior and activities in the training room. By identifying the identity-related feature vectors of different students, monitoring functions such as student entry and exit records and student activity trajectories can be realized, providing more data support for training room management.

[0031] In particular, the fusion of the semantic global feature vector of the training room course information and the student identity association feature vector can combine the course information and student identity information to achieve more comprehensive and accurate student identity identification and association, and provide more in-depth data analysis and decision support for the training room management system. By integrating the course information features and student identity features, a more comprehensive student identity feature representation can be established. This fused feature vector not only contains the student's basic identity information such as name and student number, but also includes more detailed features such as the courses the student participates in, learning status, interests and hobbies. Such a fused feature vector can more comprehensively describe the student's activities and behaviors in the training room, and provide more dimensional information for the identification and association of student identities.

[0032] Figure 3 FIG. 1 is a block diagram of a training room monitoring image feature extraction unit in an artificial intelligence-based training room management platform according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the training room monitoring image feature extraction unit 121 includes: a training room monitoring feature encoding subunit 1211, which is used to perform feature encoding on the training room monitoring image collected by the camera to obtain a training room monitoring significant feature map; a training room monitoring feature pooling subunit 1212, which is used to perform global mean pooling on the training room monitoring significant feature map to obtain a training room monitoring significant feature vector; a training room monitoring feature optimization subunit 1213, which is used to perform depth correlation compensation of the position scattering response of the training room monitoring significant feature vector in the class divergence space based on the training room student identity fusion feature vector to obtain the optimized training room monitoring significant feature vector.

[0033] It should be understood that by encoding the features of surveillance images, the complex information in the images can be converted into feature vectors with semantic meaning. These feature vectors can capture the key features in the images, such as people, objects, scenes, etc., so as to understand and analyze the content of surveillance images. By analyzing the surveillance salient feature map, managers can find problems in time, optimize the layout, and make more scientific management decisions.

[0034] Furthermore, the practice of globally mean pooling the significant feature map of the training room monitoring to obtain the significant feature vector of the training room monitoring is intended to compress the rich information in the image into a more representative feature vector. Here, by performing mean value pooling on the significant feature map, the image features can be converted from a two-dimensional matrix form to a one-dimensional feature vector. This processing method helps to reduce the dimension of the data and reduce the computational complexity, while retaining the main information in the image, making the feature vector more interpretable and operational. In this way, input images of different sizes can obtain consistent feature representations during the feature extraction process. This unified feature representation helps to improve the generalization ability of the model, so that the model can effectively extract and recognize features for images of different sizes and scales.

[0035] Furthermore, integrating the identity of students in the training room into the feature vector can provide richer contextual information for the monitoring image. Student identity information can include student identification codes, course information, etc. By integrating this information with the monitoring image features, we can better understand the behavior and activities of students in the monitoring scene and achieve a more accurate description of the internal situation of the training room. Through the deep correlation compensation of the position scattering response in the class divergence space, the position relationship and depth information between different categories in the monitoring image can be effectively processed. This compensation method can help the model better understand the correlation between different categories in the monitoring scene, thereby improving the representation ability and discrimination of the monitoring image features. In addition, in the process of optimizing the significant feature vector of the training room monitoring, the deep correlation compensation can help the model learn a more robust and generalized feature representation. By considering the deep correlation between different features, the important information and features in the monitoring image can be better captured, and the model's recognition and understanding ability of the training room monitoring scene can be improved.

[0036] In a specific embodiment of the present application, the training room monitoring feature encoding subunit 1211 includes: performing target detection on the training room monitoring image captured by the camera to obtain the training room monitoring area of ​​interest; passing the training room monitoring area of ​​interest through the training room monitoring significant target feature encoder to obtain the training room monitoring significant feature map.

[0037] It should be understood that target detection can help identify various target objects in monitoring images, such as people, objects, vehicles, etc., so as to determine the activities and situations in the training room. By accurately identifying these targets, it is possible to achieve status monitoring and real-time analysis inside the training room, which helps to detect abnormal situations and improve safety. Target detection can help extract the areas of interest in the training room monitoring images, reducing the computational complexity and data volume of subsequent processing. In addition, by locating and extracting the target areas of interest, you can focus on more in-depth analysis and processing of these areas, thereby more effectively realizing the utilization and management of the training room monitoring data.

[0038] Furthermore, the training room monitoring region of interest is processed through the training room monitoring salient target feature encoder, aiming to extract key features and encode them into more representative and distinguishable feature maps, so as to achieve a deeper understanding and analysis of the training room monitoring scene. Through the extraction of the training room monitoring region of interest, attention can be focused on the analysis of the actual area of ​​concern, reducing the interference of irrelevant information and improving processing efficiency and accuracy. This helps to more accurately capture key targets such as students, teachers or other important objects in complex monitoring scenes. This includes extracting and representing features such as shape, texture, color, etc. of the target, thereby converting the original image information into a feature map with more semantic and representational capabilities. Specifically, each layer of the convolutional neural network model including the salient target detector performs the following operations on the input data in the forward pass of the layer: convolution processing is performed on the input data using a first convolution kernel to obtain a first convolution feature map; convolution processing is performed on the first convolution feature map using a second convolution kernel to obtain a second convolution feature map, wherein the size of the first convolution kernel is larger than the size of the second convolution kernel; and, pooling processing is performed on the second convolution feature map to obtain a pooled feature map; and, activation processing is performed on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the training room monitoring salient target feature encoder is the training room monitoring salient feature map, and the input of the first layer of the training room monitoring salient target feature encoder is the training room monitoring region of interest.

[0039] In a specific embodiment of the present application, the training room monitoring feature optimization subunit 1213 includes: calculating the positional subtraction of the training room monitoring significant feature vector and the training room student identity fusion feature vector to obtain a phase difference feature vector; dividing the eigenvalues ​​of each position of the phase difference feature vector by two and then taking the absolute value to calculate the natural exponential function value to obtain an exponential feature vector; dividing the second hyperparameter by each eigenvalue of the exponential feature vector to obtain a second weighting coefficient vector; multiplying the second weighting coefficient vector by the training room student identity fusion feature vector by position to obtain a weighted training room student identity fusion feature vector; multiplying the first predetermined hyperparameter by the training room monitoring significant feature vector by position and then adding it to the weighted training room student identity fusion feature vector by position to obtain the optimized training room monitoring significant feature vector.

[0040] It should be understood that in the technical solution of the present application, it is considered that in the process of extracting the significant feature vector of the training room monitoring and the feature vector of the student identity fusion of the training room, there may be some shared information or features. For example, the significant feature vector of the training room monitoring and the feature vector associated with the student identity may contain the location information, behavioral characteristics, etc. of the relevant students, and these information may be repeated in the two feature vectors. In the process of feature extraction, sometimes different feature extractors may extract similar or even identical features. For example, if the encoder used to extract the significant features of the training room monitoring and the encoder used to extract the student identity features have overlapping feature extraction capabilities in some aspects, then there will be repeated or redundant parts in the extracted feature vector. There may be some information intersection between the significant feature vector of the training room monitoring and the feature vector of the student identity fusion. For example, if the behavior of the student in the training room affects the target detection result in the monitoring image, then the student's identity information and the target information in the monitoring image will overlap to a certain extent in the feature vector. Based on this, in the technical solution of the present application, based on the training room student identity fusion feature vector, the depth correlation compensation of the position scattering response in the class divergence space is performed on the training room monitoring significant feature vector to eliminate the repeated or redundant parts in the training room monitoring significant feature vector and the training room student identity fusion feature vector, thereby improving the generalization ability of the training room monitoring significant feature vector.

[0041] Specifically, based on the training room student identity fusion feature vector, the depth correlation compensation of the position scattering response in the divergence space is performed on the training room monitoring significant feature vector to obtain the optimized training room monitoring significant feature vector, including: based on the training room student identity fusion feature vector, the depth correlation compensation of the position scattering response in the divergence space is performed on the training room monitoring significant feature vector according to the following optimization formula to obtain the optimized training room monitoring significant feature vector;

[0042]

[0043] in, represents the eigenvalue of the ith position of the significant eigenvector monitored in the training room, represents the eigenvalue of the i-th position of the training room student identity fusion feature vector, α and ∈ represent the first predetermined hyperparameter and the second predetermined hyperparameter, ⊙ represents multiplication by position, It means adding by position. Represents the eigenvalue of the ith position of the significant eigenvector monitored by the optimization training room.

[0044] In order to eliminate the repeated or redundant parts in the training room monitoring significant feature vector and the training room student identity fusion feature vector, so as to improve the generalization ability of the training room monitoring significant feature vector, in the technical solution of the present application, based on the training room student identity fusion feature vector, the training room monitoring significant feature vector is subjected to the depth correlation compensation of the position scattering response in the class divergence space. The depth of the position scattering response of the training room monitoring significant feature vector and the training room student identity fusion feature vector in the class divergence space is simulated by calculating the gradient of the variance between the position by position of the training room monitoring significant feature vector and the training room student identity fusion feature vector, and the gradient of the variance is used to construct a depth correlation compensation factor to perform position-by-position feature depth correlation compensation on the training room monitoring significant feature vector. In this way, the repeated or redundant parts in the training room monitoring significant feature vector and the training room student identity fusion feature vector are suppressed to improve the generalization ability of the training room monitoring significant feature vector.

[0045] In a specific embodiment of the present application, the training room course information feature extraction unit 122 includes: passing the training room course information collected from the database through a training room course information semantic feature encoder to obtain multiple training room course information feature vectors; splicing the multiple training room course information feature vectors to obtain the training room course information semantic global feature vector.

[0046] It should be understood that the application of the training room course information semantic feature encoder can convert the text description, keywords, time arrangement and other contents in the course information into numerical feature vectors. This encoding method can better express the semantic information of the course, so that different courses can be compared and matched. Through the generation of the training room course information feature vector, the effective representation and storage of the course information can be achieved. Encoding the course information in the form of a feature vector can reduce the redundancy and dimension of the data and improve the compactness and expression ability of the information. This helps to efficiently store and process a large amount of course information in the training room management system, and provide users with fast and accurate query and retrieval functions. Specifically, the training room course information collected by the database is segmented to obtain a word sequence; the embedding layer of the training room course information semantic feature encoder is used to map each word in the word sequence to a word embedding vector to obtain a sequence of word embedding vectors; the converter-based Bert model of the training room course information semantic feature encoder is used to perform global context semantic encoding on the sequence of the word embedding vector to obtain multiple training room course information feature vectors.

[0047] Furthermore, by splicing the feature vectors of multiple training room course information, the feature information of each course can be integrated into a global and comprehensive feature vector to more comprehensively characterize the course information in the training room management system and provide richer information support for the decision-making and recommendation tasks of the system. In this way, the feature information of each course can be organically combined to form a more comprehensive and integrated feature representation. This global feature vector can cover the key features of multiple courses, such as subject, difficulty, type, etc., so as to better capture the feature distribution and structure of the entire course information space, and provide the system with a deeper understanding and analysis capability of the course.

[0048] In a specific embodiment of the present application, the student identity information feature extraction unit 123 includes: constructing the student identity information collected by the RFID reader installed in the training room into multiple student identity input vectors; passing the multiple student identity input vectors through a student identity feature encoder based on a convolutional neural network to obtain the student identity associated feature vector.

[0049] It should be understood that by constructing multiple student identity input vectors, the identity information of each student can be converted into an independent numerical vector representation. This representation method can include the student's identity information in a structured vector, so that the system can more conveniently identify, distinguish and manage students. Among them, each student identity input vector can include the student's ID, name, grade, major and other information, providing the system with comprehensive and accurate student identity features. In this way, it can help the system to uniformly process and manage student identity information. By converting student identity information into vector form, a standardized representation of student information can be achieved, avoiding duplication and redundancy of information, and improving the efficiency of information utilization. This helps the system better manage the identity information of a large number of students and ensure the consistency and accuracy of the information.

[0050] Furthermore, multiple student identity input vectors can be effectively learned and represented by a deep learning model through a student identity feature encoder based on a convolutional neural network, thereby providing a more representative and interpretable student identity feature representation for the training room management system. Among them, the student identity feature encoder based on a convolutional neural network can effectively capture the spatial and hierarchical features in the student identity information. Through the layer-by-layer feature extraction structure of the convolutional neural network, the system can learn the local patterns and global structures in the student identity information, so as to better understand the complexity and diversity of the student identity information. The student identity association feature vector obtained by the student identity feature encoder can help the system better understand the relationship and mutual influence between students. The convolutional neural network can take into account the correlation between student identity information when learning feature representation, thereby generating an association feature vector that can reflect the similarities and differences between students. Specifically, each layer of the student identity feature encoder based on the convolutional neural network is used to perform convolution processing, mean pooling processing based on the local feature matrix, and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the student identity feature encoder based on the convolutional neural network outputs the student identity associated feature vector, wherein the input of the student identity feature encoder based on the convolutional neural network is the multiple student identity input vectors.

[0051] In the above-mentioned training room management platform 100 based on artificial intelligence, the suspicious person early warning judgment module 130 is used to determine whether to issue a suspicious person early warning alarm based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector. It should be understood that by optimizing the training room monitoring significant feature vector, the system can extract key visual features from the monitoring image, such as the behavior, posture, and object carrying of the personnel. These significant feature vectors can help the system identify abnormal behaviors or abnormal situations, such as potential dangerous behaviors, unusual activity patterns, etc., and provide important clues for suspicious personnel identification. The training room student identity fusion feature vector contains the identity information and feature representation of each student, such as the student's ID, name, grade, major, etc. By fusing the student identity information with the monitoring image features, it is possible to track and analyze the behavior and activities of specific students in the monitoring scene. This fusion feature vector can help the system associate the behavior in the monitoring image with the identity of a specific student, and more accurately identify students who may have security risks or abnormal situations. By comprehensively analyzing and optimizing the significant feature vectors of training room monitoring and the fusion feature vectors of student identities in the training room, the system can use the deep learning model to effectively learn and model these two aspects of information, thereby realizing intelligent identification and early warning of suspicious persons in the training room. In addition, the system can quickly and accurately determine situations where there may be security risks based on the comprehensive analysis of student identity information and monitoring image features, and issue corresponding early warning alarms to help managers take timely measures to ensure the safety and order of the training room. This intelligent early warning system for suspicious persons can improve the efficiency and safety of training room management and provide a safer and more comfortable learning environment for students and faculty and staff.

[0052] Figure 4 FIG. 1 is a block diagram of a suspicious personnel early warning judgment module in an artificial intelligence-based training room management platform according to an embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, the suspicious person early warning judgment module 130 includes: a suspicious person feature fusion unit 131, which is used to associate the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector to obtain a suspicious person early warning judgment feature vector; a warning alarm judgment generation unit 132, which is used to pass the suspicious person early warning judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a suspicious person early warning alarm.

[0053] It should be understood that by associating the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector, the system can comprehensively consider the visual features and student identity information in the monitoring image, thereby obtaining a more comprehensive and comprehensive feature representation. This suspicious person early warning judgment feature vector can better reflect the details and characteristics of student behavior in the monitoring scene, help the system to more accurately judge potential security risks and abnormal situations, issue early warning alarms in time, and provide important decision-making support for managers. Specifically, the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector are fused with the following cascade formula to obtain a suspicious person early warning judgment feature vector;

[0054] Wherein, the cascade formula is:

[0055] V=Concat[V1,V2]

[0056] Among them, V1 represents the significant feature vector of the optimized training room monitoring, V2 represents the fusion feature vector of the student identity of the training room, V represents the warning judgment feature vector of the suspicious person, and Concat[·,·] represents the cascade function.

[0057] Furthermore, the suspicious person warning judgment feature vector is obtained by comprehensively optimizing the significant feature vector of the training room monitoring and the fusion feature vector of the training room student identity, which contains the associated representation of the monitoring image features and the student identity information. This feature vector is input into the classifier, and the machine learning algorithm can be used to analyze and classify it to determine whether the situation represented by the feature vector meets the characteristics of the suspicious person. This automatic warning system based on machine learning can effectively reduce the workload of managers, improve the efficiency and safety of training room management, and create a safer and more comfortable learning environment for students and faculty and staff.

[0058] In summary, the embodiment of the present application first obtains the training room monitoring image captured by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room, and then uses deep learning technology to perform feature extraction and association analysis on the three. Finally, a classifier is used to determine whether to issue a suspicious person warning alarm to help managers better understand the operation of the training room, thereby strengthening the safety prevention capabilities of the training room and improving the overall safety level of the training room.

[0059] Furthermore, in another embodiment of the present application, the artificial intelligence-based training room management platform also has a natural language recognition function, which collects voice data through a microphone installed in the training room by integrating natural language processing (NLP) technology. Then, the voice is converted into text data using voice recognition technology. Text analysis technology, such as sentiment analysis, keyword extraction, etc., is then used to analyze the converted text to identify possible abnormal behavior or inappropriate remarks. Finally, combined with the training room monitoring images and voice recognition results, a comprehensive analysis is performed through the suspicious personnel early warning judgment module, and a corresponding early warning is issued.

[0060] In another embodiment of the present application, the artificial intelligence-based training room management platform also includes learning and examination functions. First, before the training, the necessary course learning materials and online learning environment are provided through the platform. By integrating the online examination system, students are required to complete course learning and pass the examination before conducting experiments. The examination system may include functions such as question bank management, online answering, and automatic scoring. The examination results are associated with student identity information, and only students who pass the examination can obtain experimental permissions to ensure experimental safety.

[0061] In another embodiment of the present application, the training room management platform based on artificial intelligence also includes experimental result recording and knowledge point accumulation functions. It allows students to input or upload experimental data and results after the experiment by designing an experimental result recording system. The platform automatically collects and stores these data to form an experimental file for each student. The experimental result data is analyzed using machine learning technology to identify common errors and success cases and form a knowledge point library. The knowledge point library is combined with the course content to provide students with personalized learning suggestions and improvement measures. Through systematic recording and accumulation of knowledge points, students and teachers are helped to understand the key points and potential risks in the experimental process and improve the quality of experimental teaching.

[0062] As described above, the artificial intelligence-based training room management platform 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the artificial intelligence-based training room management platform 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the artificial intelligence-based training room management platform 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the artificial intelligence-based training room management platform 100 can also be one of the many hardware modules of the terminal device.

[0063] Alternatively, in another example, the artificial intelligence-based training room management platform 100 and the terminal device may also be separate devices, and the artificial intelligence-based training room management platform 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0064] Figure 5 FIG. 1 is a flow chart of a training room management method based on artificial intelligence according to an embodiment of the present application. Figure 5 As shown, according to the artificial intelligence-based training room management method of the embodiment of the present application, it includes: S110, acquiring the training room monitoring image captured by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; S120, extracting the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector from the training room monitoring image captured by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; S130, based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector, judging whether to issue a suspicious person early warning alarm.

[0065] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned training room management method based on artificial intelligence have been referred to above. Figures 1 to 4 The description of the artificial intelligence-based training room management platform has been introduced in detail, and therefore, its repeated description will be omitted.

[0066] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application.

[0067] like Figure 6 As shown, the electronic device 10 includes an input device 11, an input interface 12, a central processing unit 13, a memory 14, an output interface 15, an output device 16, and a bus 17. The input interface 12, the central processing unit 13, the memory 14, and the output interface 15 are connected to each other through the bus 17, and the input device 11 and the output device 16 are connected to the bus 17 through the input interface 12 and the output interface 15, respectively, and then connected to other components of the electronic device 10.

[0068] Specifically, the input device 11 receives input information from the outside and transmits the input information to the central processing unit 13 through the input interface 12; the central processing unit 13 processes the input information based on the computer executable instructions stored in the memory 14 to generate output information, temporarily or permanently stores the output information in the memory 14, and then transmits the output information to the output device 16 through the output interface 15; the output device 16 outputs the output information to the outside of the electronic device 10 for user use.

[0069] In one embodiment, Figure 6 The electronic device 10 shown can be implemented as a network device, which may include: a memory configured to store programs; a processor configured to run the programs stored in the memory to execute any one of the artificial intelligence-based training room management methods described in the above embodiments.

[0070] According to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network, and / or installed from a removable storage medium.

[0071] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0072] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present application, but the present application is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of the present application, and these modifications and improvements are also considered to be within the scope of protection of the present application.

Claims

1. A training room management platform based on artificial intelligence, characterized in that: include: The training room management data acquisition module is used to obtain the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; A training room management data extraction module is used to extract optimized training room monitoring significant feature vectors and training room student identity fusion feature vectors from the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; A suspicious person early warning judgment module is used to judge whether to issue a suspicious person early warning alarm based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector; Wherein, the training room management data extraction module includes: A training room monitoring image feature extraction unit, used for extracting features from the training room monitoring image collected by the camera to obtain the optimized training room monitoring significant feature vector; A training room course information feature extraction unit, used for extracting features from the training room course information collected from the database to obtain a semantic global feature vector of the training room course information; A student identity information feature extraction unit, used for extracting features from the student identity information collected by the RFID reader / writer installed in the training room to obtain a student identity-related feature vector; A student identity feature fusion unit, used for fusing the semantic global feature vector of the training room course information and the student identity associated feature vector to obtain the training room student identity fusion feature vector; Wherein, the training room monitoring image feature extraction unit includes: A training room monitoring feature coding subunit, used for performing feature coding on the training room monitoring image collected by the camera to obtain a training room monitoring significant feature map; A training room monitoring feature pooling subunit, used for performing global mean pooling on the training room monitoring significant feature map to obtain a training room monitoring significant feature vector; A training room monitoring feature optimization subunit is used to perform depth correlation compensation of position scattering response in class divergence space on the training room monitoring significant feature vector based on the training room student identity fusion feature vector to obtain the optimized training room monitoring significant feature vector; Wherein, the training room monitoring feature coding subunit includes: Performing target detection on the training room monitoring image captured by the camera to obtain a training room monitoring area of ​​interest; The training room monitoring region of interest is passed through a training room monitoring significant target feature encoder to obtain the training room monitoring significant feature map; Wherein, the training room monitoring feature optimization subunit includes: Calculate the positional subtraction of the training room monitoring significant feature vector and the training room student identity fusion feature vector to obtain a difference feature vector; Dividing the eigenvalues ​​of each position of the phase difference eigenvector by two and taking the absolute value, and then calculating the natural exponential function value to obtain an exponential eigenvector; Dividing a second hyperparameter by each eigenvalue of the exponentially modified eigenvector to obtain a second weighting coefficient vector; Multiplying the second weighting coefficient vector by the training room student identity fusion feature vector by position to obtain a weighted training room student identity fusion feature vector; The first predetermined hyperparameter is multiplied by the training room monitoring significant feature vector by position, and then added to the weighted training room student identity fusion feature vector by position to obtain the optimized training room monitoring significant feature vector.

2. The artificial intelligence-based training room management platform according to claim 1 is characterized in that: The training room course information feature extraction unit comprises: Passing the training room course information collected from the database through a training room course information semantic feature encoder to obtain a plurality of training room course information feature vectors; The multiple training room course information feature vectors are concatenated to obtain the training room course information semantic global feature vector.

3. The artificial intelligence-based training room management platform according to claim 2 is characterized in that: The student identity information feature extraction unit comprises: The student identity information collected by the RFID reader / writer installed in the training room is constructed into a plurality of student identity input vectors; The multiple student identity input vectors are passed through a student identity feature encoder based on a convolutional neural network to obtain the student identity associated feature vector.

4. The artificial intelligence-based training room management platform according to claim 3 is characterized in that: The suspicious person early warning judgment module includes: A suspicious person feature fusion unit, used to associate the optimized training room monitoring significant feature vector with the training room student identity fusion feature vector to obtain a suspicious person early warning judgment feature vector; The early warning alarm judgment generating unit is used to pass the suspicious person early warning judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a suspicious person early warning alarm.

5. A training room management method based on artificial intelligence, using the training room management platform based on artificial intelligence according to claim 1, characterized in that: include: Obtain the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader installed in the training room; Extracting optimized training room monitoring significant feature vectors and training room student identity fusion feature vectors from the training room monitoring images collected by the camera, the course information collected by the database, and the student identity information collected by the RFID reader / writer installed in the training room; Based on the optimized training room monitoring significant feature vector and the training room student identity fusion feature vector, it is determined whether to issue a suspicious person early warning alarm.

Images