Intelligent monitoring system for college students' class state

Abstract

The application discloses a kind of university student class state intelligent monitoring system, it is related to intelligent monitoring application technical field, contains sitting posture pressure monitoring device, visual state monitoring device, state reminding device, communication module and intelligent control device, the intelligent control device with main control unit as core, the information transmission of sitting posture pressure monitoring device and visual state monitoring device is given main control unit by communication module, the real-time class state of student is obtained by main control unit by intelligent analysis, abnormal control state reminding device is stimulated to remind in different degree, and data transmission is carried out to open source hardware APP.The application is combined by sitting posture pressure monitoring and visual pupil tracking state monitoring Multi-modal sensing scheme, accurate class state monitoring result is obtained by intelligent analysis, and through tactile reminder and camera flash reminder, automation, private instant intervention is realized, by using open source hardware and localization intelligent processing, system cost is effectively controlled.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring application technology, specifically to an intelligent monitoring system for university students' classroom attendance. Background Technology

[0002] University students' academic life differs from that of high school students. High school students are supervised and disciplined by teachers throughout their studies, while university students are relatively more relaxed, and their classroom behavior is also more undisciplined. They frequently fidget, and some even daydream, sleep, or use their phones, which affects the quality of teaching for instructors and also impacts the learning of other students. Therefore, many universities have attempted to design monitoring systems to track students' classroom participation, but these systems have the following drawbacks: I. Technical and accuracy bottlenecks: frequent data distortion and misjudgments.

[0003] The system is sensitive to environmental interference. Common situations such as changes in classroom lighting, students wearing masks / hats, reflections from glasses, and students in the front row blocking the view of those in the back row can cause the system to make misjudgments, resulting in a large amount of error in the raw data collected.

[0004] The rigid definition of behavior is the core technical flaw. The system simplistically quantifies the complex state of "focus" as "facing the podium" and "looking up," leading to many misjudgments. It misjudges attentive students as distracted; for example, when a student is diligently taking notes, reading a textbook, or thinking about a problem, the system will classify it as "looking down," "playing on a phone," or "unfocused." Conversely, it misjudges distracted students as attentive; for example, a student's eyes are fixed on the podium, but their mind is elsewhere, yet the system will still label them as "focused." Furthermore, the system fails to recognize "effective interaction," such as students engaging in group discussions, smiling knowingly when they understand, or frowning when they are puzzled—active learning behaviors that the system cannot identify and may even classify as "whispering" or "fidgeting."

[0005] Second, the core flaw of "can only detect, cannot intervene": the system is disconnected from the teaching process.

[0006] This is the most fatal problem in the current system, as it creates a broken management loop: For students: Lack of immediate feedback and absence of an early warning mechanism. The system detects students being distracted or using their phones, but fails to issue any immediate or friendly reminders (such as screen pop-ups or slight vibration alerts). Students are completely unaware that they have been "marked" by the system, and therefore cannot adjust themselves, missing the optimal opportunity for education and correction.

[0007] For teachers: It distracts from teaching and is too costly to intervene. Teachers cannot afford to be distracted by monitoring; their primary responsibility is to deliver lessons smoothly, observe students' genuine reactions, and interact with them. Expecting teachers to simultaneously lecture and keep a close eye on data and alerts on a monitoring screen is impractical and counterproductive, severely disrupting the continuity of classroom teaching and the teacher's focus.

[0008] Interventions are awkward and ineffective: even if teachers receive an alert that "a student has been distracted for 10 minutes", publicly reminding them will interrupt the class and hurt the student's self-esteem; after-class talks are ineffective due to the passage of time, which makes it impossible to transform the test data into effective teaching actions, but instead becomes a psychological burden for teachers.

[0009] Third, the economic costs and returns are out of balance: the input-output ratio is extremely low.

[0010] High initial investment: It requires the deployment of high-performance cameras, edge computing devices, servers, and the purchase or custom development of professional software systems, which is a huge expense.

[0011] Continuous maintenance and upgrade costs: The system requires dedicated personnel for maintenance, the algorithm model needs continuous optimization to adapt to new situations (such as new mobile phone models), and the software needs to be upgraded, all of which will bring long-term economic burdens.

[0012] The returns are unclear: a huge financial investment yields a data report whose accuracy is questionable, which cannot be directly intervened in, and which offers limited guidance for improving teaching. From the fundamental goal of improving classroom teaching quality, this investment has extremely low cost-effectiveness. Schools could have used this money for more direct and effective purposes such as teacher training, curriculum development, or improving the teaching environment. Summary of the Invention

[0013] To address the aforementioned technical problems, this invention provides an intelligent monitoring system for university students' classroom status. It employs a multimodal sensing scheme combining posture pressure monitoring and visual pupil tracking monitoring. Through intelligent analysis, it obtains accurate monitoring results of classroom status and achieves automated, private, and real-time intervention via tactile and camera flash alerts. By utilizing open-source hardware and localized intelligent processing, system costs are effectively controlled.

[0014] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A smart monitoring system for university students' classroom attendance includes: A posture pressure monitoring device is installed at the bottom of each chair to determine the student's class status by monitoring the continuous changes in the center of gravity of the sitting posture. A visual state monitoring device is installed directly above each seat. It captures pupil images of students' faces under a near-infrared light source through visual components and uses a convolutional neural network to determine the students' class status. A status reminder device, which, when the monitoring device detects an abnormal student's class status, reminds the student to adjust to a normal class status through external stimuli. The communication module enables efficient communication between the main control unit and various information acquisition and execution devices via wired or wireless means, allowing for real-time monitoring and feedback. The intelligent control device, with a main control unit as its core, transmits information from the posture pressure monitoring device and the visual state monitoring device to the main control unit through a communication module. The main control unit obtains the student's real-time class status through intelligent analysis, and controls the status reminder device to provide different levels of stimulation reminders when abnormalities occur, and transmits the data to the open-source hardware APP.

[0015] The sitting pressure monitoring device includes a seat cushion, strain gauges, and a signal conditioning circuit. The seat cushion uses highly elastic leather as the sensitive interface. Several strain gauges are arranged in a matrix with orthogonal arrangement on the leather base. Each intersection point forms a pressure-sensitive unit. When the user sits down, the weight pressure causes the strain gauges to change resistance. The resistance of each strain gauge is connected to the signal conditioning circuit, which converts the resistance change signal into a standard voltage signal. This signal is then transmitted in real time to the main control unit through distributed data acquisition nodes, enabling visualization of the seat pressure distribution for intelligent analysis by the main control unit.

[0016] The posture pressure monitoring device distinguishes students' classroom status based on the following criteria: prolonged stillness - possibly sleeping; frequent changes in posture - focused listening; significant backward or sideways leaning - distracted.

[0017] The signal conditioning circuit uses a Wheatstone bridge circuit.

[0018] The visual state monitoring device uses an OpenMV camera chip and implements the core machine vision algorithm in C language to achieve tasks such as color block detection, face detection, eye tracking, edge detection, and marker tracking. It also has built-in support for CNN convolutional neural networks. Through hierarchical feature learning, it can autonomously identify various behavioral features such as looking down, looking up, looking around, and using a mobile phone. The OpenMV camera chip is connected to the main control unit, which uploads the real-time collected data and uploads it to the open-source App through the communication module.

[0019] The state judgment algorithm of the visual state monitoring device is set as follows: Listen attentively: Pupils are aligned with the direction of the podium / blackboard, blinking frequency is normal, and posture is upright; Inattentiveness / distraction: Unfocused gaze, prolonged fixation on unrelated areas, possibly accompanied by frequent nodding, dozing off, or stiffness; Sleep: Eyes closed beyond the set threshold, head drooping, sitting posture data shows prolonged stillness; When using a mobile phone: Keep your gaze focused downwards on the hand area. At this time, you can try to use visual aids to identify the rectangular outline of the phone for assistance.

[0020] The status reminder device includes a light flashing stimulation device and a seat local stimulation device. The light flashing stimulation device is implemented through the visual component of the visual status monitoring device. When the device detects that a student is daydreaming, distracted, or playing with a mobile phone, it reminds the student by controlling the light flashing stimulation of the student's pupils through the visual component. The seat local stimulation device provides a slight reminder when the visual component detects that a student is not paying attention, is distracted, and is leaning back or to the side significantly. It provides a strong reminder when the visual component detects that a student is sleeping, playing with a mobile phone, or is sitting still for a long time.

[0021] The seat local stimulation device includes a rotating motor, a worm gear, and a pin. The rotating motors are respectively located on the side and front of the seat. Each rotating motor is connected to a worm gear and a pin. The rotating motor is connected to a microcontroller, which is connected to a main control unit. The microcontroller receives instructions to control the displacement of the pin. For a slight reminder, the pin is pushed out 2mm. For a strong reminder, the pin is pushed out 5mm three times.

[0022] The main control unit uses an STM32F427 CPU. By integrating the pressure distribution data from the posture pressure monitoring device and the visual posture information from the visual state monitoring device, it can achieve intelligent analysis of posture, pressure distribution balance, and student class status, providing data support for applications such as healthy posture reminders and pressure distribution optimization.

[0023] The control method of the intelligent monitoring system for university students' classroom attendance is implemented through the following steps: 1) The strain gauge matrix and the OpenMV camera synchronously collect sitting pressure data and pupil and head posture information. This data is uploaded to the main control unit through NB communication protocol and asynchronous serial communication. After CNN model and data fusion analysis, the control unit determines whether the student is listening attentively, daydreaming, sleeping or playing on their mobile phone, and then sends control commands to the corresponding slave station. 2) When the system determines that a student is in a "distracted / inattentive" state for 3 consecutive seconds, and the posture detection shows a significant backward / sideways lean, a mild reminder process is triggered: Flashing reminder: The CPU sends a flash command to the OpenMV camera via the UART interface. The camera uses strobe infrared technology to emit a soft flash, which stimulates the student's pupils and reminds them to concentrate. The flashing stops automatically after 1 second. Tactile alert: The CPU sends a preset pulse signal to the rotary motor driver, which drives the worm gear to rotate and drives the worm wheel to rotate. The rotational motion is converted into linear motion of the ejector pin through the eccentric wheel. The ejector pin extends precisely by 2mm to achieve a slight tactile alert. After 0.5 seconds, the ejector pin automatically retracts. 3) When the system determines that a student is in a "sleeping" or "playing on their phone" state for more than 5 seconds, a strong reminder process is triggered: Flash alert: The CPU sends a high-frequency flash command to the OpenMV camera via the UART interface. The camera then activates a high-frequency flash mode and flashes continuously for 3 seconds, enhancing the alert effect through multiple pupil stimulations. Tactile cues: The CPU sends a strong pulse signal to the rotating motor, controlling the ejector pin to extend precisely by 5mm, hold for 1 second, and then retract. This process is repeated every 0.5 seconds, completing 3 consecutive ejection-retraction actions to ensure a strong yet harmless tactile cues for students. The beneficial effects of this invention are as follows: 1. A multimodal data fusion intelligent monitoring system was designed. It achieves contactless monitoring through pressure sensors and cameras, and performs intelligent analysis by combining pressure distribution and visual information to improve the accuracy of judging students' class status and overcome the problem of inaccurate results from single visual monitoring. 2. Two status reminder devices, tactile reminder and camera flash reminder, were designed to provide mild and obvious stimulation according to the different states of students, so as to realize the immediate intervention of abnormal states. This intervention is only for the individual students with abnormal states, which is highly private and does not affect the classroom order, reducing interference to teachers and other students, and solving the core pain point of "only detecting without intervention". 3. An intelligent control device was designed with a modular design, which facilitates the addition of other sensors or functions. The main control unit is the core of the device, which intelligently analyzes the monitoring results of students' class status and transmits the data and monitoring results to open source hardware. This combines open source hardware with local intelligent processing, eliminating the need for continuous optimization and system upgrades, effectively controlling system costs, and transforming intelligent monitoring from "passive monitoring" to "active assistance". This truly serves to improve the quality of classroom teaching and is suitable for large-scale promotion and application in universities. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the system structure principle of the present invention; Figure 2 This is a schematic diagram of the structure of the seat local stimulation device of the present invention. Detailed Implementation

[0025] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments: like Figure 1As shown, the intelligent monitoring system for college students' classroom status includes a posture pressure monitoring device 1, a visual status monitoring device 2, a status reminder device 3, a communication module 4, and an intelligent control device 5.

[0026] The sitting posture pressure monitoring device 1 is installed at the bottom of each seat. By monitoring the continuous changes in the center of gravity of the sitting posture, it can determine the student's class status. The visual state monitoring device 2 is installed directly above each seat. It captures pupil images of the student's face under a near-infrared light source through visual components and uses a convolutional neural network to determine the student's class status. When the monitoring device detects an abnormal student's class status, the status reminder device 3 reminds the student to adjust to a normal class status through external stimuli. The communication module 4 enables efficient communication between the main control unit and each information acquisition device and execution device via wired or wireless means, and provides real-time monitoring and feedback. The intelligent control device 5 is centered on the main control unit 51. It transmits information from the sitting posture pressure monitoring device 1 and the visual state monitoring device 2 to the main control unit 51 through the communication module 4. The main control unit 51 obtains the student's real-time class status through intelligent analysis. When there is an abnormality, it controls the status reminder device 3 to provide different levels of stimulation reminders and transmits the data to the open-source hardware APP 6.

[0027] In this preferred embodiment, the sitting posture pressure monitoring device 1 includes a seat cushion, strain gauges, and a signal conditioning circuit. The seat cushion uses highly elastic leather as the sensitive interface. A matrix of strain gauges arranged orthogonally on the leather base forms a pressure-sensitive unit at each intersection. When the user sits down, the weight pressure causes a change in the resistance of the strain gauges. The resistance of each strain gauge is connected to the signal conditioning circuit, which converts the resistance change signal into a standard voltage signal. This signal is then transmitted in real-time to the main control unit 51 through distributed data acquisition nodes, enabling visualization of the seat pressure distribution. Simultaneously, an OpenMV visual sensor installed above the seat captures the user's posture image. The two images are combined for intelligent analysis by the main control unit 51.

[0028] Specifically, the posture pressure monitoring device 1 distinguishes students' class status based on the following criteria: prolonged stillness - possibly sleeping; frequent changes in posture - attentive listening; significant backward or sideways lean - distracted.

[0029] Preferably, the signal conditioning circuit uses a Wheatstone bridge circuit.

[0030] As a preferred embodiment, the visual state monitoring device 2 uses an OpenMV camera chip 21, and implements the core machine vision algorithm in C language to achieve tasks such as color block detection, face detection, eye tracking, edge detection, and marker tracking. It also has built-in support for CNN convolutional neural networks, including face detection, head pose estimation, eye state recognition, and gaze direction estimation. Through hierarchical feature learning, it can autonomously identify various behavioral features such as looking down, looking up, looking around, and using a mobile phone. The OpenMV camera chip 21 is connected to the main control unit 51, which uploads the real-time collected data and uploads it to the open-source App 6 through the communication module 4.

[0031] Convolutional Neural Networks (CNNs) are deep learning models specifically designed for processing image data. CNNs excel in detecting student classroom behavior due to their powerful feature extraction and adaptive learning capabilities. Because students' postures vary greatly in the classroom environment, CNNs can autonomously identify various behavioral features such as looking down, looking up, looking around, and using mobile phones through hierarchical feature learning. Furthermore, CNNs' parameter sharing mechanism and local connectivity characteristics enable them to efficiently extract features when processing complex background images and perform well when training on large datasets, thereby improving the model's accuracy and robustness. Specifically, the state judgment algorithm of the visual state monitoring device 2 is set as follows: Listen attentively: Pupils are aligned with the direction of the podium / blackboard, blinking frequency is normal, and posture is upright; Inattentiveness / distraction: Unfocused gaze, prolonged fixation on unrelated areas, possibly accompanied by frequent nodding, dozing off, or stiffness; Sleep: Eyes closed beyond the set threshold, head drooping, sitting posture data shows prolonged stillness; When using a mobile phone: Keep your gaze focused downwards on the hand area. At this time, you can try to use visual aids to identify the rectangular outline of the phone for assistance.

[0032] In a preferred embodiment, the status reminder device 3 includes a light strobe stimulation device 31 and a seat local stimulation device 32. The light strobe stimulation device 31 is implemented through the visual component of the visual status monitoring device 2. When the device detects that a student is daydreaming, distracted, or playing with their phone, it reminds the student by controlling the light strobe stimulation of the student's pupils through the visual component. In this embodiment, the camera of the visual component uses strobe infrared technology, and the camera is controlled to flash according to the monitoring results.

[0033] The seat local stimulation device 32 provides a slight reminder when the visual component detects that the student is not paying attention, is distracted, or is leaning back or to the side significantly. It also provides a strong reminder when the visual component detects that the student is sleeping, playing on their phone, or is motionless for an extended period of time.

[0034] Specifically, in this embodiment, as Figure 2 As shown, the seat local stimulation device 32 includes a rotating motor 321, a worm gear 322, and a pin 323. The rotating motors 321 are respectively disposed on the side and front end of the seat. Each rotating motor 321 is connected to a worm gear 322 and a pin 323. The rotating motors 321 are connected to a microcontroller 324, which is connected to a main control unit 51. When the microcontroller 324 receives instructions from the main control unit 51, it sends a series of pulse signals to the driver of the rotating motors 321. The output shaft of the rotating motors 321 directly drives the worm gear to rotate, which in turn drives the worm gear to rotate. When the worm gear drives the eccentric wheel to rotate, the center of the eccentric wheel makes a circular motion around the axis. Due to the eccentric effect, it will continuously push and pull the pin 323 in contact with it in the vertical direction, thereby converting the rotational motion into the linear reciprocating motion of the pin 323. The number of pulses sent from the main control unit 51 to the rotating motor 321 can precisely control the final rotation angle of the worm gear, thereby precisely controlling the extension displacement of the ejector pin 323. For example, when there is a slight reminder, the ejector pin is controlled to extend out 2mm, and when there is a strong reminder, the ejector pin is controlled to extend out 5mm three times in a row.

[0035] As a preferred embodiment, the main control unit 51 adopts an STM32F427 CPU. By integrating the pressure distribution data of the sitting posture pressure monitoring device 1 with the visual posture information of the visual state monitoring device 2, it realizes intelligent analysis of sitting posture, pressure distribution balance, and student's class status, providing data support for applications such as healthy sitting posture reminders and pressure distribution optimization.

[0036] The communication module 4 used in this invention can employ wireless communication such as Bluetooth, Wi-Fi, and wired serial communication, depending on the actual situation. A custom data frame format is used, including a frame header, data length, command word, and data content, to ensure reliable transmission. If the number of slave stations is large, ZigBee or WiFi modules could be considered, but the nRF24L01 is low-cost and easy to implement; therefore, the nRF24L01 is used in this embodiment.

[0037] The control method of the intelligent monitoring system for university students' classroom attendance is implemented through the following steps: 1) The strain gauge matrix and the OpenMV camera synchronously collect sitting pressure data and pupil and head posture information. This data is uploaded to the main control unit through NB communication protocol and asynchronous serial communication. After CNN model and data fusion analysis, the control unit determines whether the student is listening attentively, daydreaming, sleeping or playing on their mobile phone, and then sends control commands to the corresponding slave station. 2) When the system determines that a student is in a "distracted / inattentive" state for 3 consecutive seconds, and the posture detection shows a significant backward / sideways lean, a mild reminder process is triggered: Flashing reminder: The CPU sends a flash command to the OpenMV camera via the UART interface. The camera uses strobe infrared technology to emit a soft flash, which stimulates the student's pupils and reminds them to concentrate. The flashing stops automatically after 1 second. Tactile alert: The CPU sends a preset pulse signal to the rotary motor driver, which drives the worm gear to rotate and drives the worm wheel to rotate. The rotational motion is converted into linear motion of the ejector pin through the eccentric wheel. The ejector pin extends precisely by 2mm to achieve a slight tactile alert. After 0.5 seconds, the ejector pin automatically retracts. 3) When the system determines that a student is in a "sleeping" or "playing on their phone" state for more than 5 seconds, a strong reminder process is triggered: Flash alert: The CPU sends a high-frequency flash command to the OpenMV camera via the UART interface. The camera then activates a high-frequency flash mode and flashes continuously for 3 seconds, enhancing the alert effect through multiple pupil stimulations. Tactile cues: The CPU sends a strong pulse signal to the rotating motor, controlling the ejector pin to extend precisely by 5mm, hold for 1 second, and then retract. This process is repeated every 0.5 seconds, completing 3 consecutive ejection-retraction actions to ensure a strong yet harmless tactile cues for students. Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A smart monitoring system for university students' classroom attendance, characterized in that, Includes: A sitting posture pressure monitoring device (1) is installed at the bottom of each seat to monitor the continuous changes in the center of gravity of the sitting posture and determine the student's class status. Visual state monitoring device (2), the visual state monitoring device is set directly above each seat, and captures the pupil image of the student's face under near-infrared light source through visual components, and uses convolutional neural network to judge the student's class status; Status reminder device (3), when the monitoring device detects that the student's class status is abnormal, the status reminder device reminds the student to adjust to a normal class status through external stimulation; The communication module (4) enables efficient communication between the main control unit and each information acquisition device and execution device via wired or wireless means, and provides real-time monitoring and feedback. The intelligent control device (5) is based on the main control unit (51). The communication module (4) transmits the information of the sitting pressure monitoring device (1) and the visual state monitoring device (2) to the main control unit (51). The main control unit (51) obtains the student's real-time class status through intelligent analysis. When there is an abnormality, it controls the status reminder device to provide different levels of stimulation reminders and transmits the data to the open-source hardware APP (6).

2. The intelligent monitoring system for university students' class attendance as described in claim 1, characterized in that, The sitting pressure monitoring device (1) includes a seat cushion, strain gauges and a signal conditioning circuit. The seat cushion uses high-elasticity leather as the sensitive interface. Several strain gauges arranged in a vertical and horizontal orthogonal layout are set on the leather base to form a matrix. Each intersection point forms a pressure sensitive unit. When the user sits down, the weight pressure causes the strain gauges to generate resistance changes. The resistance of each strain gauge is connected to the signal conditioning circuit, which converts the resistance change signal into a standard voltage signal and transmits it to the main control unit (51) in real time through distributed data acquisition nodes to realize the visualization of the seat pressure distribution for intelligent analysis by the main control unit (51).

3. The intelligent monitoring system for university students' class attendance according to claim 2, characterized in that, The sitting posture pressure monitoring device (1) distinguishes students' class status based on the following criteria: prolonged stillness - possibly sleeping; frequent adjustment of sitting posture - focused listening; large backward or sideways lean - distracted.

4. The intelligent monitoring system for university students' class attendance as described in claim 2, characterized in that, The signal conditioning circuit uses a Wheatstone bridge circuit.

5. The intelligent monitoring system for university students' class attendance according to claim 1, characterized in that, The visual state monitoring device (2) uses an OpenMV camera chip (21) and implements the core machine vision algorithm in C language to achieve color block search, face detection, eye tracking, edge detection, and sign tracking. It also has built-in support for CNN convolutional neural networks. Through hierarchical feature learning, it can autonomously identify various behavioral features such as looking down, looking up, looking around, and using a mobile phone. The OpenMV camera chip is connected to the main control unit (51) to upload the data collected in real time and upload it to the open source App (6) through the communication module (4).

6. The intelligent monitoring system for university students' class attendance according to claim 5, characterized in that, The state judgment algorithm of the visual state monitoring device (2) is set as follows: Listen attentively: Pupils are aligned with the direction of the podium / blackboard, blinking frequency is normal, and posture is upright; Inattentiveness / distraction: Unfocused gaze, prolonged fixation on unrelated areas, possibly accompanied by frequent nodding, dozing off, or stiffness; Sleep: Eyes closed beyond the set threshold, head drooping, sitting posture data shows prolonged stillness; When using a mobile phone: Keep your gaze focused downwards on the hand area. At this time, you can try to use visual aids to identify the rectangular outline of the phone for assistance.

7. The intelligent monitoring system for university students' class attendance according to claim 1, characterized in that, The status reminder device (3) includes a light flashing stimulation device (31) and a seat local stimulation device (32). The light flashing stimulation device (31) is implemented through the visual component of the visual status monitoring device (2). When the device detects that a student is daydreaming, distracted, or playing with a mobile phone, it will remind the student by controlling the light flashing stimulation of the visual component to stimulate the student's pupils. The seat local stimulation device (32) will provide a slight reminder when the visual component detects that the student is not serious, distracted, and leans back or to the side in a large way. It will also provide a strong reminder when the visual component detects that the student is sleeping, playing with a mobile phone, or sitting still for a long time.

8. The intelligent monitoring system for university students' class attendance according to claim 7, characterized in that, The seat local stimulation device (32) includes a rotating motor (321), a worm gear (322), and a pin (323). The rotating motor (321) is respectively located on the side and front of the seat. Each rotating motor is connected to a worm gear (322) and a pin (323). The rotating motor (321) is connected to a microcontroller (324). The microcontroller (324) is connected to a main control unit (51) and receives instructions to control the displacement of the pin (323). When a slight reminder is given, the pin (323) is pushed out 2mm. When a strong reminder is given, the pin (323) is pushed out 5mm three times.

9. The intelligent monitoring system for university students' class attendance according to claim 1, characterized in that, The main control unit (51) adopts an STM32F427 CPU. By integrating the pressure distribution data of the sitting posture pressure monitoring device (1) with the visual posture information of the visual state monitoring device (2), it realizes intelligent analysis of sitting posture, pressure distribution balance and student class status, and provides data support for applications such as healthy sitting posture reminders and pressure distribution optimization.

10. The control method of the intelligent monitoring system for college students' class attendance as described in claim 1, characterized in that, This can be achieved through the following steps: 1) The strain gauge matrix and the OpenMV camera synchronously collect sitting pressure data and pupil and head posture information. This data is uploaded to the main control unit through NB communication protocol and asynchronous serial communication. After CNN model and data fusion analysis, the control unit determines whether the student is listening attentively, daydreaming, sleeping or playing on their mobile phone, and then sends control commands to the corresponding slave station. 2) When the system determines that a student is in a "distracted / inattentive" state for 3 consecutive seconds, and the posture detection shows a significant backward / sideways lean, a mild reminder process is triggered: Flashing reminder: The CPU sends a flash command to the OpenMV camera via the UART interface. The camera uses strobe infrared technology to emit a soft flash, which stimulates the student's pupils and reminds them to concentrate. The flashing stops automatically after 1 second. Tactile alert: The CPU sends a preset pulse signal to the rotary motor driver, which drives the worm gear to rotate and drives the worm wheel to rotate. The rotational motion is converted into linear motion of the ejector pin through the eccentric wheel. The ejector pin extends precisely by 2mm to achieve a slight tactile alert. After 0.5 seconds, the ejector pin automatically retracts. 3) When the system determines that a student is in a "sleeping" or "playing on their phone" state for more than 5 seconds, a strong reminder process is triggered: Flash alert: The CPU sends a high-frequency flash command to the OpenMV camera via the UART interface. The camera then activates a high-frequency flash mode and flashes continuously for 3 seconds, enhancing the alert effect through multiple pupil stimulations. Tactile cues: The CPU sends a strong pulse signal to the rotating motor to control the pin to extend precisely by 5mm, hold for 1 second, and then retract. This process is repeated every 0.5 seconds to complete 3 consecutive push-out and retraction actions, ensuring a strong tactile cues effect that does not harm the student.

Images