Machine vision-based intelligent attendance system for student guarding sentry

Abstract

The invention discloses a machine vision-based intelligent attendance system for student guarding posts, which relates to the field of image analysis, and comprises an integrated control module which is used as a central management and control terminal and is used for editing and operating operation indexes of each functional module and receiving and displaying feedback data of each module; the illumination auxiliary module is used as an adjustable light source to cooperatively work with the light sensor, monitors the environment illumination intensity of the preset school gate area in real time, and automatically adjusts the illumination intensity according to the environment illumination intensity; the image acquisition module is used for capturing a real-time video stream of a school gate area in real time through an industrial camera; identity verification, on-duty state judgment and equipment compliance inspection are deeply integrated, and non-inductive and non-contact synchronous identification and verification are performed on personnel such as security personnel, traffic police and parents and volunteers on duty during rush hours when students go to and go to school, so that the manpower management cost is greatly saved, mistakes, omissions and disputes caused by human factors are avoided, and the safety of students is improved. And the traceability of the whole process is ensured by continuous data flow.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, specifically to a machine vision-based intelligent attendance system for school nursing staff. Background Technology

[0002] With the education authorities strengthening campus security measures, school safety patrols play a crucial role in maintaining order at school gates and preventing potential risks. Traffic congestion and frequent student movement in front of schools, along with external environmental factors such as weather changes and lighting conditions, can increase safety hazards. School safety patrol personnel must be on duty on time and equipped with standard equipment, such as riot helmets and stab-proof vests, to protect students' safety to and from school. With the increasing prevalence of artificial intelligence and machine vision in the security field, industrial cameras and edge computing devices are beginning to enter public areas such as school gates, enabling real-time monitoring and data processing, filling the intelligent gaps in traditional security systems.

[0003] Traditional attendance methods rely on manual recording and patrols, which makes the attendance process time-consuming and labor-intensive. They cannot achieve real-time, all-weather monitoring, lack the ability to simultaneously identify multiple targets, and make it difficult for manual or simple monitoring systems to judge the status of personnel in a timely manner, resulting in attendance data delays or errors, increasing security risks. They cannot maintain stable performance in complex environments such as rain, backlight, or obstruction, and are sensitive to weather changes or uneven lighting, and cannot adjust automatically, resulting in poor image acquisition and recognition effects. Summary of the Invention

[0004] (a) Technical problems to be solved In view of the above-mentioned shortcomings of the existing technology, the present invention provides an intelligent attendance system for school nursing posts based on machine vision, which can effectively solve the problems of the existing technology.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: This invention discloses a machine vision-based intelligent attendance system for school nursing staff, comprising: The integrated control module, as the central management and control terminal, edits and runs the operating indicators of each functional module, and receives and displays the feedback data from each module; The lighting assistance module, as an adjustable light source, works in conjunction with the light sensor to monitor the ambient light intensity of the preset school gate area in real time and automatically adjust the light intensity according to the ambient light intensity. The image acquisition module is used to capture real-time video streams of the school gate area using an industrial camera. It acquires multi-target optical images including security personnel, traffic police, and parent volunteers at a preset frame rate and resolution, and converts the optical signals into analog electrical signals. The signal conversion module is used to convert analog electrical signals into digital image signals through the image acquisition card, select the best transmission path, and output the generated digital image. The feature extraction module is used to process digital images based on edge detection and image segmentation algorithms, locate target areas, and extract facial biometrics, equipment outlines, and spatial pose feature vectors of security personnel. The target recognition unit is used to input the extracted feature information into the pre-trained recognition model and output the recognition results of identity, equipment and attendance. The judgment and analysis module is used to comprehensively judge the recognition results of the target recognition unit by the module. Combining the preset allowance threshold and model confidence, it analyzes whether the target meets the requirements. If it does not meet the requirements, an alarm signal is generated. The data storage module is used to perform desensitization processing on biometric data in attendance records.

[0006] Furthermore, the target recognition unit is further equipped with sub-modules, including an identity recognition module, an equipment detection module, and an attendance association module, wherein: The identity recognition module is used to obtain the feature vector output by the feature extraction module through a pre-trained recognition model. It first detects the facial region in the image, extracts the facial feature embedding vector, and then compares it with the features in the registration database. If the similarity exceeds the preset threshold, it is determined to be a legitimate on-duty personnel. It supports the simultaneous recognition of multiple people's identities. The equipment detection module is used to locate riot helmets, stab-proof vests, riot shields, and riot forks using a multi-target detection algorithm. If the detection confidence of all four types of equipment is greater than a preset threshold, the equipment is deemed compliant; otherwise, a violation record is generated and missing items are marked. The attendance association module is used to spatially and temporally associate the identity recognition results of the identity recognition module with the equipment detection results of the equipment detection module. When a person appears in the preset attendance area within the specified time period and both their identity and equipment status are compliant, an on-duty compliance signal is output. Otherwise, a real-time abnormal alarm is triggered, and independent attendance statistics for multiple people are supported.

[0007] Furthermore, the expression for calculating similarity in the identity recognition module is as follows: ; In the formula, S represents the similarity score between the facial feature to be identified and the registered feature, and its value ranges from [−1, 1]. This represents the facial feature embedding vector extracted from the real-time image. This represents the facial feature embedding vector of registered individuals pre-stored in the database.

[0008] Furthermore, the signal conversion module is interconnected with an image preprocessing module via a wireless network. The image preprocessing module is used to perform noise reduction filtering, geometric distortion correction, and contrast enhancement operations on the digital image. It uses median filtering to eliminate random noise, affine transformation to correct lens distortion, and histogram equalization to process the image feature recognizability.

[0009] Furthermore, the adjustable light source in the lighting assistance module is a number of LED units with multi-level brightness controllable. The adjustment logic is as follows: when the light sensor detects that the ambient light intensity is lower than a preset threshold, the light intensity compensation is dynamically calculated based on the difference; when the ambient light intensity is higher than the preset threshold, the directional supplementary lighting mode is activated, the LED units in the background area are turned off, and only the local LED array facing the target person is activated for lateral supplementary lighting.

[0010] Furthermore, the light sensor in the light-assisted module adopts the following deployment strategy: it collects light data at each point in real time at the four corners and the center point of the preset school gate area, and uses the average weighted value of each sensor reading as the criterion for ambient light intensity. The weight allocation is dynamically adjusted according to a preset threshold based on the straight-line distance between the sensor and the target person's activity area.

[0011] Furthermore, the process of selecting the best transmission path in the signal conversion module includes: monitoring the delay, bandwidth and packet loss rate of each transmission channel in real time through the dynamic path evaluation module, and selecting the path with the highest comprehensive transmission quality score; when the quality score of the primary path is lower than the preset disaster recovery threshold, automatically switching to the backup wireless transmission channel and maintaining the continuous output of digital image signals.

[0012] Furthermore, when the judgment and analysis module makes a comprehensive judgment on the recognition results, it receives the identity recognition confidence level and equipment detection confidence level from the target recognition unit, compares the identity recognition confidence level with a preset identity allowance threshold, and compares the equipment detection confidence level with a preset equipment allowance threshold. If the identity recognition confidence level is lower than the identity allowance threshold or the equipment detection confidence level is lower than the equipment allowance threshold, the target is determined to be non-compliant and an alarm signal is generated.

[0013] Furthermore, when performing the desensitization process, the data storage module applies a blurring algorithm to the facial area of ​​the personnel in the attendance record for visual masking, extracts the feature vector generated from the facial biometric data, and applies an encryption algorithm to the feature vector for encrypted storage.

[0014] Furthermore, the integrated control module is interconnected with the illumination assist module, the image acquisition module, and the data storage module via a wireless network; the signal conversion module is interconnected with the image acquisition module and the feature extraction module via a wireless network; and the target recognition unit is interconnected with the feature extraction module and the judgment and analysis module via a wireless network.

[0015] (III) Beneficial Effects Compared with known prior art, the technical solution provided by this invention has the following beneficial effects: 1. By deeply integrating identity verification, on-duty status judgment, and equipment compliance inspection through machine vision technology, multiple people, including security personnel, traffic police, and parent volunteers, can be identified and verified in a seamless and non-contact manner during peak hours when students are going to and from school. This greatly saves manpower management costs, avoids errors and disputes caused by human factors, and ensures full traceability through continuous data flow, thereby improving security.

[0016] 2. Through ambient lighting control and image preprocessing algorithms, the system effectively overcomes the effects of adverse environmental factors such as rain, backlighting, and occlusion, ensuring that effective images can be acquired and stably analyzed under different weather and lighting conditions. It can be adapted to existing edge computing devices such as IPC cameras at school gates, without the need to deploy expensive large servers, resulting in low implementation costs and low response latency.

[0017] 3. By performing real-time desensitization and encrypted storage on the collected biometric information, only necessary non-identity attendance records are retained, fundamentally eliminating the risk of personal information leakage and misuse. This not only improves the security level of the campus perimeter but also incorporates a built-in privacy protection mechanism, achieving a balance between security and compliance. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0019] Figure 1 This is a schematic diagram of the overall framework of the present invention; Figure 2 This is a schematic diagram of the target recognition unit in this invention.

[0020] The numbers in the diagram represent: 1. Integrated control module; 2. Illumination auxiliary module; 3. Image acquisition module; 4. Signal conversion module; 5. Feature extraction module; 6. Target recognition unit; 61. Identity recognition module; 62. Equipment detection module; 63. Attendance association module; 7. Judgment and analysis module; 8. Data storage module; 9. Image preprocessing module. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] The present invention will be further described below with reference to embodiments.

[0023] This embodiment of the intelligent attendance system for school safety posts based on machine vision, such as... Figure 1 and Figure 2 As shown, it includes: Integrated control module 1, as the central control terminal, edits and runs the operating indicators of each functional module, and receives and displays the feedback data from each module; The illumination assist module 2, as an adjustable light source, works in conjunction with the light sensor to monitor the ambient light intensity of the preset school gate area in real time and automatically adjust the light intensity according to the ambient light intensity; ensuring the image clarity of the target object in rainy or backlit scenes; Image acquisition module 3 is used to capture real-time video streams of the school gate area in real time using an industrial camera, such as a CCD or CMOS camera, to acquire multi-target optical images including security personnel, traffic police and parent volunteers at a preset frame rate and resolution, and to convert optical signals into analog electrical signals. Signal conversion module 4 is used to convert analog electrical signals into digital image signals through an image acquisition card, select the best transmission path, and output the generated digital image. Signal conversion module 4 is interconnected with image preprocessing module 9 via a wireless network. Image preprocessing module 9 is used to perform noise reduction filtering, geometric distortion correction, and contrast enhancement operations on the digital image. It uses median filtering to eliminate random noise, affine transformation to correct lens distortion, and histogram equalization to process the image feature recognizability. The process of selecting the best transmission path includes: real-time monitoring of the delay, bandwidth, and packet loss rate of each transmission channel through a dynamic path evaluation module, and selecting the path with the highest comprehensive transmission quality score; when the quality score of the primary path is lower than the preset disaster recovery threshold, it automatically switches to the backup wireless transmission channel and maintains continuous output of digital image signals.

[0024] Feature extraction module 5 is used to process digital images based on edge detection and image segmentation algorithms, locate target areas, and extract facial biometrics, equipment outlines, and spatial pose feature vectors of security personnel. The target recognition unit 6 is used to input the extracted feature information into a pre-trained recognition model and output the identification results of identity, equipment, and attendance. The target recognition unit 6 has sub-modules, including an identity recognition module 61, an equipment detection module 62, and an attendance association module 63, wherein: The identity recognition module 61 is used to obtain the feature vector output by the feature extraction module 5 through the pre-trained recognition model. First, it detects the facial region in the image, extracts the facial feature embedding vector, and then compares the similarity with the features in the registration database. If the similarity exceeds the preset threshold, it is determined to be a legitimate on-duty personnel. It supports the simultaneous recognition of multiple identities. The equipment detection module 62 is used to locate riot helmets, stab-proof vests, riot shields and riot forks through a multi-target detection algorithm. If the detection confidence of the four types of equipment is greater than the preset threshold, the equipment is judged to be compliant; otherwise, a violation record is generated and the missing item is marked. The attendance association module 63 is used to spatiotemporally associate the identity recognition result of the identity recognition module 61 with the equipment detection result of the equipment detection module 62. When a person appears in the preset attendance area within the specified time period and both the identity and equipment status are compliant, an on-duty compliance signal is output; otherwise, a real-time abnormal alarm is triggered, and independent attendance statistics for multiple people are supported.

[0025] The judgment and analysis module 7 is used to comprehensively judge the recognition results of the target recognition unit 6, and analyze whether the target meets the requirements by combining the preset allowable threshold and model confidence. If it does not meet the requirements, an alarm signal is generated. When the judgment and analysis module 7 comprehensively judges the recognition results, it receives the identity recognition confidence and equipment detection confidence from the target recognition unit 6, compares the identity recognition confidence with the preset identity allowable threshold, and compares the equipment detection confidence with the preset equipment allowable threshold. If the identity recognition confidence is lower than the identity allowable threshold or the equipment detection confidence is lower than the equipment allowable threshold, the target is determined to not meet the requirements and an alarm signal is generated.

[0026] Data storage module 8 is used to perform desensitization processing on biometric data in attendance records. During the desensitization process, a blurring algorithm is applied to the facial area of ​​the personnel in the attendance records for visual masking, feature vectors generated from facial biometric data are extracted, and encryption algorithms are applied to the feature vectors for encrypted storage.

[0027] In one embodiment of this invention, the integrated control module 1 is interconnected with the illumination assist module 2, the image acquisition module 3, and the data storage module 8 via a wireless network; the signal conversion module 4 is interconnected with the image acquisition module 3 and the feature extraction module 5 via a wireless network; and the target recognition unit 6 is interconnected with the feature extraction module 5 and the judgment and analysis module 7 via a wireless network.

[0028] Compared with existing technologies, this technology significantly improves attendance efficiency and reduces human error through self-ambient lighting optimization and dynamic transmission path switching mechanisms. It integrates advanced image processing, feature extraction, and recognition mechanisms to effectively cope with complex scenarios, improve recognition accuracy and anti-interference capabilities, and achieves comprehensive judgment of identity, equipment, and attendance through data desensitization and encryption, forming a safer and more intelligent closed-loop management system. This reduces overall operating costs and enhances campus security response speed.

[0029] At other levels, this embodiment also provides a representation for calculating similarity, specifically: ; In the formula, S represents the similarity score between the facial feature to be identified and the registered feature, and its value ranges from [−1, 1]. This represents the facial feature embedding vector extracted from the real-time image. This represents the facial feature embedding vector of registered individuals pre-stored in the database.

[0030] This embodiment provides an adjustable light source, specifically a plurality of LED units with multi-level controllable brightness. The adjustment logic is as follows: when the light sensor detects that the ambient light intensity is lower than a preset threshold, the light intensity compensation is dynamically calculated based on the difference; when the ambient light intensity is higher than the preset threshold, the directional supplementary lighting mode is activated, the LED units in the background area are turned off, and only the local LED array facing the target person is activated for lateral supplementary lighting; the light sensor adopts the following deployment strategy: it collects light data at each point in real time at the four corners and the center point of the preset school gate area, and uses the average weighted value of the readings of each sensor as the criterion for ambient light intensity. The weight allocation is dynamically adjusted according to a preset threshold based on the straight-line distance between the sensor and the target person's activity area.

[0031] Compared with existing technologies, it adopts multi-level brightness controllable LED units combined with intelligent adjustment logic, which can dynamically calculate and compensate for light intensity based on the data collected in real time by the light sensor, and achieve precise directional supplemental lighting. This effectively avoids the problem of overexposure or underexposure of images, and improves the imaging clarity of the system in rainy or backlit scenes. By deploying sensors at the four corners and center points of the preset school gate area and performing weighted average calculation based on the distance to the target person, the accuracy of the lighting criteria is improved and unnecessary energy consumption is reduced.

[0032] In summary, this invention, through efficient control and data feedback of each functional module, ensures real-time monitoring and automatic illumination adjustment, improving image quality in complex environments. Dynamic path evaluation optimizes signal transmission, guaranteeing the continuity and stability of image data. Combined with edge detection and multi-target recognition algorithms, it can simultaneously and accurately detect identity and equipment, and output attendance compliance signals in real time. Upon detection of absenteeism, missing equipment, or abnormal situations, it can immediately trigger alarms and notify administrators, proactively preventing security risks and improving the accuracy and efficiency of attendance tracking. The de-identification mechanism ensures the privacy and security of biometric data, complies with relevant laws and regulations, and reduces the risk of data leakage.

[0033] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A machine vision-based intelligent attendance system for school safety patrols, characterized in that: include: The integrated control module, as the central management and control terminal, edits and runs the operating indicators of each functional module, and receives and displays the feedback data from each module; The lighting assistance module, as an adjustable light source, works in conjunction with the light sensor to monitor the ambient light intensity of the preset school gate area in real time and automatically adjust the light intensity according to the ambient light intensity. The image acquisition module is used to capture real-time video streams of the school gate area using an industrial camera. It acquires multi-target optical images containing security personnel, traffic police, and parent volunteers at a preset frame rate and resolution, and converts the optical signals into analog electrical signals. The signal conversion module is used to convert analog electrical signals into digital image signals through the image acquisition card, select the best transmission path, and output the generated digital image. The feature extraction module is used to process digital images based on edge detection and image segmentation algorithms, locate target areas, and extract facial biometrics, equipment outlines, and spatial pose feature vectors of security personnel. The target recognition unit is used to input the extracted feature information into the pre-trained recognition model and output the recognition results of identity, equipment and attendance. The judgment and analysis module is used to comprehensively judge the recognition results of the target recognition unit by the module. Combining the preset allowance threshold and model confidence, it analyzes whether the target meets the requirements. If it does not meet the requirements, an alarm signal is generated. The data storage module is used to perform desensitization processing on biometric data in attendance records.

2. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, The target recognition unit has sub-modules deployed below it, including an identity recognition module, an equipment detection module, and an attendance association module, wherein: The identity recognition module is used to obtain the feature vector output by the feature extraction module through a pre-trained recognition model. It first detects the facial region in the image, extracts the facial feature embedding vector, and then compares it with the features in the registration database. If the similarity exceeds the preset threshold, it is determined to be a legitimate on-duty personnel. It supports the simultaneous recognition of multiple people's identities. The equipment detection module is used to locate riot helmets, stab-proof vests, riot shields, and riot forks using a multi-target detection algorithm. If the detection confidence of all four types of equipment is greater than a preset threshold, the equipment is deemed compliant; otherwise, a violation record is generated and missing items are marked. The attendance association module is used to spatially and temporally associate the identity recognition results of the identity recognition module with the equipment detection results of the equipment detection module. When a person appears in the preset attendance area within the specified time period and both their identity and equipment status are compliant, an on-duty compliance signal is output. Otherwise, a real-time abnormal alarm is triggered, and independent attendance statistics for multiple people are supported.

3. The intelligent attendance system for school safety posts based on machine vision according to claim 2, characterized in that, The expression for calculating similarity in the identity recognition module is as follows: ； In the formula, S represents the similarity score between the facial feature to be identified and the registered feature, and its value ranges from [−1, 1]. This represents the facial feature embedding vector extracted from the real-time image. This represents the facial feature embedding vector of registered individuals pre-stored in the database.

4. The intelligent attendance system for school safety posts based on machine vision according to claim 1, characterized in that, The signal conversion module is interconnected with the image preprocessing module via a wireless network. The image preprocessing module is used to perform noise reduction filtering, geometric distortion correction and contrast enhancement operations on the digital image. It uses median filtering to eliminate random noise, affine transformation to correct lens distortion, and histogram equalization to process the image feature recognizability.

5. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, The adjustable light source in the lighting assistance module consists of several LED units with multi-level controllable brightness. The adjustment logic is as follows: when the light sensor detects that the ambient light intensity is lower than a preset threshold, the light intensity compensation is dynamically calculated based on the difference; when the ambient light intensity is higher than the preset threshold, the directional supplementary lighting mode is activated, the LED units in the background area are turned off, and only the local LED array facing the target person is activated for lateral supplementary lighting.

6. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, The light sensor in the light-assisted module adopts the following deployment strategy: it collects light data at each point in real time at the four corners and the center point of the preset school gate area, and uses the average weighted value of each sensor reading as the criterion for ambient light intensity. The weight allocation is dynamically adjusted according to a preset threshold based on the straight-line distance between the sensor and the target person's activity area.

7. The intelligent attendance system for school safety posts based on machine vision according to claim 1, characterized in that, The process of selecting the best transmission path in the signal conversion module includes: real-time monitoring of the delay, bandwidth and packet loss rate of each transmission channel through the dynamic path evaluation module, and selection of the path with the highest comprehensive transmission quality score; when the quality score of the primary path is lower than the preset disaster recovery threshold, it automatically switches to the backup wireless transmission channel and maintains the continuous output of digital image signals.

8. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, When the judgment and analysis module makes a comprehensive judgment on the recognition results, it receives the identity recognition confidence level and equipment detection confidence level from the target recognition unit, compares the identity recognition confidence level with the preset identity allowance threshold, and compares the equipment detection confidence level with the preset equipment allowance threshold. If the identity recognition confidence level is lower than the identity allowance threshold or the equipment detection confidence level is lower than the equipment allowance threshold, the target is determined to be non-compliant and an alarm signal is generated.

9. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, When performing the desensitization process, the data storage module applies a blurring algorithm to the facial area of ​​the personnel in the attendance record for visual masking, extracts the feature vector generated from the facial biometric data, and applies an encryption algorithm to the feature vector for encrypted storage.

10. The intelligent attendance system for school nursing posts based on machine vision according to claim 1, characterized in that, The integrated control module is interconnected with the illumination assist module, image acquisition module, and data storage module via a wireless network. The signal conversion module is interconnected with the image acquisition module and feature extraction module via a wireless network. The target recognition unit is interconnected with the feature extraction module and judgment and analysis module via a wireless network.

Images