Real-time crowd-flow statistics method and system based on internet of things information acquisition

Abstract

The present invention provides a real-time crowd-flow statistical method and system based on Internet of Things information acquisition. The method comprises: acquiring a classroom monitoring picture by means of a camera; an electronic class board performing processing on the classroom monitoring picture, storing a processed result, and uploading the result to a local data center; and the local data center generating a real-time crowd-flow statistical result and displaying the real-time crowd-flow statistical result to teachers and students by means of a visual interface. The system corresponding to the method comprises: a monitoring module, a processing module, and a display module. By means of the method and system, a student can be helped to quickly find a vacant classroom, the resource utilization rate of classrooms can be improved, the student on-time attendance rate can be standardized, and teachers can be helped to count attendance. Moreover, monitoring resources are better utilized, avoiding unnecessary monitoring, and reducing the burden on cameras and devices for storing monitoring resources as well as on people when reviewing monitoring.

Description

Real-time crowd flow statistics method and system based on Internet of Things information collection Technical Field

[0001] The present invention proposes a real-time pedestrian flow statistics method and system based on information collection of the Internet of Things, belonging to the field of application of Internet of Things technology. Background Art

[0002] In recent years, with the continuous development of technology and the internet, many industries have a strong demand for crowd counting, such as at high-traffic locations like train stations, bus stops, subway stations, and shopping mall entrances. People counting systems enable convenient, reliable, and real-time counting of people in various locations without disrupting the public. Combined with image analysis technology, these systems can clearly and quickly capture passenger flow dynamics, providing data support and enabling decision-makers to take timely action. The primary advantage of using image processing for people counting systems is that the image signals are highly intuitive and easy to interpret.

[0003] Real-time classroom traffic flow is also crucial for university students. Today, students can access the internet anytime, anywhere, and with limited resources in university libraries and study rooms, vacant classrooms often become the primary study area for many students. However, lacking visibility into classroom resource usage, students often spend considerable time searching for classrooms with vacancies and no classes. For university teachers, attendance statistics are an effective way to document students' course progress and provide a basis for regular grade assessments. Traditional methods of assessing attendance are often time-consuming and labor-intensive, and the presence of alternate sign-ins makes the data unreliable. This wastes valuable class time and places an additional burden on teachers.

[0004] Summary of the Invention

[0005] The present invention provides a method and system for real-time crowd flow statistics based on IoT information collection to solve the above-mentioned problems:

[0006] The present invention proposes a real-time crowd flow statistics method based on Internet of Things information collection, the method comprising:

[0007] Obtain classroom surveillance images through cameras;

[0008] The electronic class board processes the classroom monitoring image, saves the processed result and uploads it to the local data center;

[0009] The local data center generates real-time crowd statistics and displays them to teachers and students through a visual interface.

[0010] Furthermore, classroom monitoring images are obtained through cameras, including:

[0011] When the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom, the electronic class sign corresponding to the classroom will query whether there is a class within the preset time period;

[0012] If the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send control signals to the cameras inside the classroom and at the classroom door. The cameras inside the classroom and at the classroom door receive the control signals, start monitoring, and send the monitoring images to the corresponding electronic class board;

[0013] If the electronic class board finds that the corresponding classroom has no classes within the preset time period, the Zigbee communication protocol is used to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

[0014] Furthermore, if the electronic class board finds that the corresponding classroom has no classes within a preset time period, it uses the Zigbee communication protocol to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring transmission frequency obtained by the frequency calculation model. The frequency calculation model is as follows: F = k1*[α*FD+(1-α)*(1-S)*Presence]

[0015] Where F is the frequency, FD represents the difference between the previous and next frames of the surveillance video, Presence is a binary variable representing the presence of a person, which is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0, α is used to balance the weight of the frame difference and the structural similarity index, k1 is a constant whose value range is [15,30] and is used to adjust the monitoring transmission frequency, S is the structural similarity index,

[0016] Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2 , where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes, I(x,y) = 1-|L x -Ly | / L max

[0017] Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

[0018] Furthermore, the electronic class board processes the classroom monitoring screen, saves the processed results and uploads them to the local data center, including:

[0019] When the electronic class card receives surveillance footage from the classroom and at the classroom door, it first compares the faces captured by the surveillance footage at the classroom door with the faces of the class that will start the class stored in the electronic class card. It then continuously uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of people. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device, who then further confirms the attendance. The teacher's confirmation information is then uploaded to the local data center.

[0020] When the electronic class card receives surveillance footage of the classroom door, it first pre-processes the data, then uses the YOLOv5s model to obtain the number of people entering and leaving the classroom, displays it through a visual interface, and then uploads it to the local data center.

[0021] During data preprocessing, the following models can be used to optimize it:

[0022] Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

[0023] Furthermore, the local data center generates real-time crowd statistics and displays them to teachers and students through a visual interface, including:

[0024] After processing the attendance information of classrooms with classes, the local data center stores it according to the attendance rate of the corresponding class hours of the course, and displays the attendance rate of each classroom just processed through a visual interface;

[0025] The local data center processes the number of people entering and leaving the vacant classrooms and arranges them according to the display format of the electronic class board, forming the overall class board display data and the class board display data for each floor of each building;

[0026] The overview shift board display data is displayed on the overview shift board, and the shift board display data of each floor of each building is transmitted to the electronic shift board device of the corresponding floor through the local area network communication mode.

[0027] The present invention proposes a real-time pedestrian flow statistics system based on Internet of Things information collection, the system comprising:

[0028] Monitoring module, used to obtain classroom monitoring images through cameras;

[0029] The processing module processes the classroom monitoring images, saves the processed results and uploads them to the local data center;

[0030] Display module: the local data center generates real-time crowd statistics results and displays them to teachers and students through a visual interface.

[0031] Furthermore, the monitoring module includes:

[0032] The module for detecting someone entering the classroom is used to check whether there is a class in the preset time period on the electronic class board corresponding to the classroom when the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom;

[0033] Start the camera module inside the classroom and at the classroom door. When the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send a control signal to the camera inside the classroom and at the classroom door. The camera inside the classroom and at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board;

[0034] Start the camera module at the classroom door. When the electronic class board finds that the corresponding classroom has no classes within the preset time period, the Zigbee communication protocol is used to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

[0035] Furthermore, the module for starting the camera at the classroom door includes a frequency calculation model module, F = k1*[α*FD+(1-α)*(1-S)*Presence]

[0036] Where F is the frequency, FD represents the difference between the previous and next frames of the surveillance video, Presence is a binary variable representing the presence of a person, which is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0, α is used to balance the weight of the frame difference and the structural similarity index, and the value range of α is (0, 1), k1 is a constant used to adjust the monitoring transmission frequency, and the value range of k1 is [15, 30], S is the structural similarity index,

[0037] Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2 , where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes, I(x,y) = 1-|L x -L y | / L max

[0038] Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

[0039] Furthermore, the processing module includes:

[0040] The class-specific processing module is used to compare the faces captured by the surveillance footage at the classroom door with the faces of the upcoming class stored in the electronic class card when the electronic class card receives surveillance footage from the classroom. It also continuously uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of heads. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device for further confirmation of the teacher's attendance. The teacher's confirmation information is then uploaded to the local data center.

[0041] The processing module for idle classrooms is used to pre-process the data when the electronic class card receives the surveillance image of the classroom door. Then, the YOLOv5s model is used to obtain the number of people entering and leaving the classroom. The data is displayed through a visual interface and then uploaded to the local data center. During the data pre-processing process, the following models can be used to optimize it.

[0042] Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

[0043] Furthermore, the display module includes:

[0044] The corresponding time attendance rate display module is used to store the attendance information of classrooms with classes after the local data center processes it according to the attendance rate of the corresponding class hours of the course, and display the attendance rate of each classroom just processed through a visual interface;

[0045] Form a display data module, which is used by the local data center to process the number of people entering and leaving the vacant classrooms, and arrange them according to the display format of the electronic class board to form the overview class board display data and the class board display data for each floor of each building;

[0046] The transmission module displays the overview class board display data on the overview class board, and transmits the class board display data of each floor of each building to the electronic class board equipment on the corresponding floor through the local area network communication method. Students can view the overview class board information by logging into the campus network to obtain the available classroom information.

[0047] The beneficial effects of the present invention are as follows: it helps students to quickly find vacant classrooms, improves the utilization rate of classroom resources, standardizes the students' on-time attendance rate and helps teachers to count attendance; it processes and analyzes classroom monitoring video images by using target detection network technology, establishes a visual interface, and provides relevant information of classroom data; it can automatically start or shut down monitoring according to the situation in the classroom, reducing unnecessary monitoring time and saving resources; through the combined use of infrared sensors and electronic class signs, the system can intelligently determine whether the classroom has classes in a specific time period, and through the automated monitoring switch, unnecessary monitoring can be avoided, thereby saving energy and reducing equipment pressure; it can provide real-time monitoring images when needed, and adjust the monitoring frequency as needed, and can effectively manage monitoring data according to specific needs. According to; because the monitoring object demand is to monitor people, the monitoring is only turned on when the infrared sensor detects that someone has entered the classroom. This not only reduces the pressure on the equipment, but also reduces the workload of people when it is necessary to review the monitoring; at the same time, when the monitoring is turned on, the monitoring needs are different when the classroom is in class and when it is not. The classroom with classes needs to take attendance and monitor the number of people in class to avoid students skipping classes. When there are no classes, the main monitoring is the flow of people in the classroom and whether there are currently vacant seats in the classroom without the need to monitor the situation in the classroom. The system can automatically switch the monitoring mode according to the class schedule and school arrangements to meet the monitoring needs in different scenarios. Through the intelligent monitoring system, schools can make better use of monitoring resources, avoid unnecessary monitoring, and reduce the burden on cameras and equipment that stores monitoring resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] FIG1 is a schematic diagram of a method for real-time crowd flow statistics based on Internet of Things information collection according to the present invention;

[0049] Figure 2 is the visualization interface of the local data center. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0051] One embodiment of the present invention provides a method for real-time crowd flow statistics based on IoT information collection, the method comprising:

[0052] Obtain classroom surveillance images through cameras;

[0053] The electronic class board processes the classroom monitoring image, saves the processed result and uploads it to the local data center;

[0054] The local data center generates real-time crowd statistics and displays them to teachers and students through a visual interface.

[0055] The working principle and effect of the above technical solution are as follows: cameras are installed in the classroom and at the classroom door to capture real-time monitoring images, which will contain information such as activities and personnel in the classroom; electronic class signs collect and process monitoring images, and detect and track personnel through computer vision technology and other image processing methods; the processed results will be saved and uploaded to the local data center through a network connection for further data storage and processing; the local data center receives the processed monitoring data, counts the entry and exit of personnel, and generates real-time crowd statistics results; finally, through a visual interface, the real-time crowd statistics results are displayed to teachers and students, allowing them to clearly understand the dynamic situation of personnel in the classroom.

[0056] One embodiment of the present invention provides a real-time crowd statistics method based on IoT information collection, which uses a camera to obtain classroom monitoring images, including:

[0057] When the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom, the electronic class sign corresponding to the classroom will query whether there is a class within the preset time period;

[0058] If the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send control signals to the cameras inside the classroom and at the classroom door. The cameras inside the classroom and at the classroom door receive the control signals, start monitoring, and send the monitoring images to the corresponding electronic class board;

[0059] If the electronic class board finds that the corresponding classroom has no classes within the preset time period, the Zigbee communication protocol is used to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

[0060] The preset time period can be set manually. Generally, a class lasts 45 minutes, so the reference preset time is 45 minutes.

[0061] The working principle of the above technical solution is: the infrared sensor installed in the classroom detects that there is no one, and the infrared sensor at the classroom door detects that someone enters the classroom; when the above situation occurs, the electronic class board will query whether there are courses scheduled for the corresponding classroom within the preset time period; send a control signal according to the course situation. If it is found that the corresponding classroom has classes within the preset time period, the electronic class board will use the Zigbee communication protocol to send control signals to the cameras in the classroom and at the classroom door, triggering the camera to start monitoring and send the monitoring screen to the electronic class board; if it is found that the corresponding classroom has no classes within the preset time period, the electronic class board will use the Zigbee communication protocol to send a control signal to the camera at the classroom door, triggering the camera to start monitoring, and send the monitoring screen to the electronic class board according to the preset monitoring frequency.

[0062] The above technical solution has the following benefits: the system can automatically activate or deactivate monitoring based on classroom conditions, reducing unnecessary monitoring time and conserving resources. By combining infrared sensors with electronic class signs, the system can intelligently determine whether a classroom is currently occupied during a specific time period. Automatic monitoring on / off prevents unnecessary monitoring, saving energy and alleviating equipment strain. The system can provide real-time monitoring footage when needed and adjust the monitoring frequency as needed, effectively managing monitoring data based on specific needs. Because the monitoring target is people, monitoring is only activated when the infrared sensor detects someone entering the classroom. This not only reduces equipment strain but also reduces the workload for personnel when reviewing the footage. Furthermore, when monitoring is activated, the monitoring requirements differ depending on whether the classroom is in class or not. Active classrooms require attendance checks and monitoring the number of people in class to prevent students from skipping. When classes are not in class, the system primarily monitors classroom traffic flow and the availability of available seats, without requiring monitoring of classroom conditions. The system can automatically switch monitoring modes based on the class schedule and school schedule to meet monitoring needs in different scenarios. Through this intelligent monitoring system, schools can better utilize monitoring resources, avoid unnecessary monitoring, and reduce the burden on cameras and the equipment storing monitoring resources.

[0063] One embodiment of the present invention provides a real-time crowd statistics method based on IoT information collection, characterized in that if the electronic class sign finds that the corresponding classroom has no classes within a preset time period, a control signal is sent to a camera at the classroom door using the Zigbee communication protocol. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class sign according to the monitoring transmission frequency obtained by a frequency calculation model. The frequency calculation model is as follows: F = k1*[α*FD+(1-α)*(1-S)*Presence]

[0064] Among them, F is the frequency, FD represents the difference between the previous frame and the next frame of the surveillance video,

[0065] Where m and n represent the height and width of the image, respectively, and A(i,j) and B(i,j) represent the pixel value at position (i,j) in images A and B, respectively.

[0066] Presence is a binary variable representing the presence of a person. It is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0. α is used to balance the weights of the frame difference and the structural similarity index. k1 is a constant used to adjust the monitoring transmission frequency, with a value range of [15,30]. S is the structural similarity index.

[0067] Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2 , where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes, I(x,y) = 1-|L x -L y | / L max

[0068] Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

[0069] The working principle and effect of the above technical solution are as follows: through this model, the system can dynamically adjust the frequency of sending monitoring images according to actual conditions, avoiding wasting bandwidth and storage resources when the number of people in the classroom has not changed; the model can reduce the frequency of sending monitoring images when there is no person or the picture changes little, thereby saving network transmission and storage resources; the weight items and parameter settings in the model make the monitoring system more intelligent and adaptable, and can dynamically adjust the monitoring frequency according to the actual scene to better meet actual needs; the calculation of F combines frame difference, structural similarity and the presence of people, so that the monitoring frequency can be dynamically adjusted according to the changes in the picture and the presence of people; the calculation of S takes into account the structural similarity and light and shadow similarity of the image, so that the frequency of the monitoring picture can be closer to the actual changes, avoiding overly frequent or sparse monitoring picture transmission; I(x,y) is a similarity function used to measure light and shadow changes, so that the model can more comprehensively consider the impact of light intensity on the monitoring frequency, because sometimes with the change of time, the change of light and shadow will also cause differences between the two photos, but at this time the monitoring frequency should not increase, because there is no Because the presence of people in the scene is important, changes in light and shadow are also factored into the monitoring transmission frequency model, further increasing the accuracy of the monitoring frequency model. The model intelligently adjusts the monitoring frequency based on scene changes, effectively avoiding the transmission and storage of invalid monitoring images. By dynamically adjusting the monitoring frequency, resource waste is avoided, improving the effectiveness and practicality of the monitoring system. Each formula comprehensively considers factors such as inter-frame differences, structural similarity, and light and shadow changes, allowing the monitoring frequency to more comprehensively and accurately reflect changes in the monitoring image. I(x,y) considers changes in light intensity and normalizes the brightness changes at the corresponding image points. Since its value range is [0,1], their differences can be easily understood and are not affected by the absolute value of light intensity. The I(x,y) model design makes the similarity function suitable for comparing images under various lighting conditions, including images captured under different lighting conditions. I(x,y) considers the maximum light intensity, so it maintains a certain degree of robustness even under different lighting conditions, thereby better reflecting the similarity between images. The above model makes the similarity calculation between images more universal and robust.

[0070] One embodiment of the present invention provides a real-time crowd flow statistics method based on Internet of Things information collection, wherein an electronic class board processes classroom monitoring images, saves the processed results, and uploads them to a local data center, including:

[0071] When the electronic class card receives surveillance footage from the classroom and at the classroom door, it first compares the faces captured by the surveillance footage at the classroom door with the faces of the class that will start the class stored in the electronic class card. It then continuously uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of people. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device, who then further confirms the attendance. The teacher's confirmation information is then uploaded to the local data center.

[0072] When the electronic class card receives surveillance footage of the classroom door, it first pre-processes the data, then uses the YOLOv5s model to obtain the number of people entering and leaving the classroom, displays it through a visual interface, and then uploads it to the local data center.

[0073] During data preprocessing, the following models can be used to optimize it:

[0074] Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

[0075] The working principle of the above technical solution is as follows: After the electronic class card receives the surveillance footage of the classroom door, it uses the pre-stored facial data of the class that will attend to compare and confirm whether the person entering the classroom is the expected one; the YOLOv5s model is used to continuously detect the number of heads in the classroom surveillance video to obtain real-time headcount data; when a face mismatch is found or the number of monitored faces is inconsistent with the actual number of heads monitored, the system will send a message to the corresponding teacher's electronic device, and the teacher will confirm the attendance. The teacher's confirmation information will be uploaded to the local data center; the surveillance footage of the classroom door is preprocessed, and the YOLOv5s model is used to obtain the number of people entering and leaving the classroom. The data is then displayed through a visual interface and uploaded to the local data center; the data is preprocessed using the morphological operation model Θ(A, B). This morphological operation model first dilates the image and then erodes the result. Through this combination, small noise can be removed while maintaining the overall shape of the target.

[0076] The effects of the above technical solution are as follows: through facial recognition and real-time headcount monitoring, the system can promptly detect whether the number of people entering the classroom meets expectations, thereby improving campus safety management. When an abnormal situation occurs, the system automatically sends a message to the corresponding teacher, who can confirm attendance and enhance the accuracy of student attendance management. Through a visual interface, the number of people entering and leaving the classroom can be intuitively displayed, providing real-time and effective data support for school management. The use of morphological operation models to pre-process data helps improve image quality and enhance the accuracy of subsequent facial recognition and headcount monitoring. At the same time, it simplifies the subsequent model processing process, can remove small noise in the image or connect some discontinuous parts, thereby more accurately identifying the target outline, and combines dilation and erosion to achieve more complex morphological transformations. Using a combination of opening and closing operations, better image processing effects are achieved.

[0077] One embodiment of the present invention provides a real-time crowd flow statistics method based on IoT information collection, wherein the local data center generates real-time crowd flow statistics and displays them to teachers and students through a visual interface, including:

[0078] After processing the attendance information of classrooms with classes, the local data center stores it according to the attendance rate of the corresponding class hours of the course, and displays the attendance rate of each classroom just processed through a visual interface;

[0079] The local data center processes the number of people entering and leaving the vacant classrooms and arranges them according to the display format of the electronic class board, forming the overall class board display data and the class board display data for each floor of each building;

[0080] The overview class board display data is displayed on the overview class board, and the class board display data of each floor of each building is transmitted to the electronic class board device on the corresponding floor through the local area network communication method. Students can view the overview class board information by logging into the campus network to obtain the available classroom information.

[0081] The working principle of the above technical solution is as follows: the local data center processes the attendance information of classrooms with classes, calculates the attendance rate of the corresponding class hours, and stores it; displays the attendance rate of each classroom through a visual interface and provides it to relevant management personnel for viewing; processes the information of idle classrooms and arranges it into the display format of electronic class boards to form overview class board display data and class board display data for each floor of each building; displays the overview class board display data on the overview class board, and then transmits the class board display data for each floor of each building to the electronic class board equipment on the corresponding floor through local area network communication; students can view the overview class board information and obtain idle classroom information by logging into the campus network.

[0082] The effects of the above technical solution are: through the processing of attendance information, school administrators can clearly understand the attendance rate of the corresponding courses in each classroom, providing data support for teaching management; the visual interface displays the attendance rate of the classroom, so that administrators can grasp the student attendance in a timely manner, providing a basis for subsequent teaching management and intervention; the vacant classroom information is arranged in the format of an electronic class board and transmitted to the electronic class board equipment on the corresponding floor, helping students to quickly obtain vacant classroom information; students can easily obtain overview class board information through the campus network, thereby quickly obtaining vacant classroom information, improving the efficiency of campus resource utilization; data transmission is completed through LAN communication, ensuring the security and immediacy of information.

[0083] One embodiment of the present invention provides a real-time pedestrian flow statistics system based on IoT information collection, the system comprising:

[0084] Monitoring module, used to obtain classroom monitoring images through cameras;

[0085] The processing module processes the classroom monitoring images, saves the processed results and uploads them to the local data center;

[0086] Display module: the local data center generates real-time crowd statistics results and displays them to teachers and students through a visual interface.

[0087] The working principle and effect of the above technical solution are as follows: cameras are installed in the classroom and at the classroom door to capture real-time monitoring images, which will contain information such as activities and personnel in the classroom; electronic class signs collect and process monitoring images, and detect and track personnel through computer vision technology and other image processing methods; the processed results will be saved and uploaded to the local data center through a network connection for further data storage and processing; the local data center receives the processed monitoring data, counts the entry and exit of personnel, and generates real-time crowd statistics results; finally, through a visual interface, the real-time crowd statistics results are displayed to teachers and students, allowing them to clearly understand the dynamic situation of personnel in the classroom.

[0088] One embodiment of the present invention provides a real-time pedestrian flow statistics system based on IoT information collection, wherein the monitoring module includes:

[0089] The module for detecting someone entering the classroom is used to check whether there is a class in the preset time period on the electronic class board corresponding to the classroom when the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom;

[0090] Start the camera module inside the classroom and at the classroom door. When the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send a control signal to the camera inside the classroom and at the classroom door. The camera inside the classroom and at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board;

[0091] Start the camera module at the classroom door. When the electronic class board finds that the corresponding classroom has no classes within the preset time period, the Zigbee communication protocol is used to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

[0092] The working principle of the above technical solution is: the infrared sensor installed in the classroom detects that there is no one, and the infrared sensor at the classroom door detects that someone enters the classroom; when the above situation occurs, the electronic class board will query whether there are courses scheduled for the corresponding classroom within the preset time period; send a control signal according to the course situation. If it is found that the corresponding classroom has classes within the preset time period, the electronic class board will use the Zigbee communication protocol to send control signals to the cameras in the classroom and at the classroom door, triggering the camera to start monitoring and send the monitoring screen to the electronic class board; if it is found that the corresponding classroom has no classes within the preset time period, the electronic class board will use the Zigbee communication protocol to send a control signal to the camera at the classroom door, triggering the camera to start monitoring, and send the monitoring screen to the electronic class board according to the preset monitoring frequency.

[0093] The above technical solution has the following benefits: the system can automatically activate or deactivate monitoring based on classroom conditions, reducing unnecessary monitoring time and conserving resources. By combining infrared sensors with electronic class signs, the system can intelligently determine whether a classroom is currently occupied during a specific time period. Automatic monitoring on / off prevents unnecessary monitoring, saving energy and alleviating equipment strain. The system can provide real-time monitoring footage when needed and adjust the monitoring frequency as needed, effectively managing monitoring data based on specific needs. Because the monitoring target is people, monitoring is only activated when the infrared sensor detects someone entering the classroom. This not only reduces equipment strain but also reduces the workload for personnel when reviewing the footage. Furthermore, when monitoring is activated, the monitoring requirements differ depending on whether the classroom is in class or not. Active classrooms require attendance checks and monitoring the number of people in class to prevent students from skipping. When classes are not in class, the system primarily monitors classroom traffic flow and the availability of available seats, without requiring monitoring of classroom conditions. The system can automatically switch monitoring modes based on the class schedule and school schedule to meet monitoring needs in different scenarios. Through this intelligent monitoring system, schools can better utilize monitoring resources, avoid unnecessary monitoring, and reduce the burden on cameras and the equipment storing monitoring resources.

[0094] One embodiment of the present invention is a real-time crowd statistics system based on IoT information collection. The module for starting the camera at the classroom door includes a frequency calculation model module, F = k1*[α*FD+(1-α)*(1-S)*Presence]

[0095] Where F is the frequency, FD represents the difference between the previous and next frames of the surveillance video, Presence is a binary variable representing the presence of a person, which is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0, α is used to balance the weight of the frame difference and the structural similarity index, and the value range of α is (0, 1), k1 is a constant used to adjust the monitoring transmission frequency, and the value range of k1 is [15, 30], S is the structural similarity index,

[0096] Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2, where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes, I(x,y) = 1-|L x -L y | / L max

[0097] Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

[0098] The working principle and effect of the above technical solution are as follows: Through this model, the system can dynamically adjust the transmission frequency of monitoring images according to actual conditions, avoiding wasting bandwidth and storage resources when the number of people in the classroom does not change. The model can reduce the transmission frequency of monitoring images when no one is present or the images do not change much, thereby saving network transmission and storage resources.

[0099] The weight items and parameter settings in the model make the monitoring system more intelligent and adaptable, and can dynamically adjust the monitoring frequency according to the actual scene to better meet actual needs; the calculation of F combines frame difference, structural similarity and the presence of people, so that the monitoring frequency can be dynamically adjusted according to changes in the picture and the presence of people; the calculation of S takes into account the structural similarity and light and shadow similarity of the image, so that the frequency of the monitoring picture can be closer to the actual changes, avoiding overly frequent or sparse monitoring pictures. I(x,y) is a similarity function used to measure the changes in light and shadow, so that the model can more comprehensively consider the impact of light intensity on the monitoring frequency, because sometimes with the change of time, the change of light and shadow will also cause the two photos to be different, but the monitoring frequency should not increase at this time because there is no person, so the light and shadow changes are also taken into account in the monitoring transmission frequency model to further increase the accuracy of the monitoring frequency model; the model intelligently adjusts the monitoring frequency according to scene changes , effectively avoiding the transmission and storage of invalid monitoring images; by dynamically adjusting the monitoring frequency, it avoids the waste of resources and improves the effectiveness and practicality of the monitoring system; each formula comprehensively considers factors such as inter-frame differences, structural similarity and light and shadow changes, so that the monitoring frequency can more comprehensively and accurately reflect the changes in the monitoring images; I(x,y) takes into account the changes in light intensity and standardizes the brightness changes on the corresponding image points. Since the value range is [0,1], their differences can be easily understood and are not affected by the absolute value of light intensity. The I(x,y) model design makes the similarity function suitable for image comparison under various lighting conditions, including images captured under different lighting conditions; I(x,y) takes into account the maximum value of light intensity, so even under different lighting conditions, it can maintain a certain robustness, thereby better reflecting the similarity between images; the above model makes the similarity calculation between images more universal and robust.

[0100] An embodiment of the present invention provides a real-time pedestrian flow statistics system based on Internet of Things information collection, wherein the processing module includes:

[0101] The class-specific processing module is used to compare the faces captured by the surveillance footage from the classroom and at the classroom door with the faces of the class that will start the class, which is stored in the electronic class card. It then uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of people in the class. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device for further confirmation of the teacher's attendance. The teacher's confirmation information is then uploaded to the local data center.

[0102] The processing module for idle classrooms is used to pre-process the data when the electronic class card receives the surveillance image of the classroom door. Then, the YOLOv5s model is used to obtain the number of people entering and leaving the classroom. The data is displayed through a visual interface and then uploaded to the local data center. During the data pre-processing process, the following models can be used to optimize it.

[0103] Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

[0104] The working principle of the above technical solution is as follows: After the electronic class card receives the surveillance footage of the classroom door, it uses the pre-stored facial data of the class that will attend to compare and confirm whether the person entering the classroom is the expected one; the YOLOv5s model is used to continuously detect the number of heads in the classroom surveillance video to obtain real-time headcount data; when a face mismatch is found or the number of monitored faces is inconsistent with the actual number of heads monitored, the system will send a message to the corresponding teacher's electronic device, and the teacher will confirm the attendance. The teacher's confirmation information will be uploaded to the local data center; the surveillance footage of the classroom door is preprocessed, and the YOLOv5s model is used to obtain the number of people entering and leaving the classroom. The data is then displayed through a visual interface and uploaded to the local data center; the data is preprocessed using the morphological operation model Θ(A, B). This morphological operation model first dilates the image and then erodes the result. Through this combination, small noise can be removed while maintaining the overall shape of the target.

[0105] The effects of the above technical solution are as follows: through facial recognition and real-time headcount monitoring, the system can promptly detect whether the number of people entering the classroom meets expectations, thereby improving campus safety management. When an abnormal situation occurs, the system automatically sends a message to the corresponding teacher, who can confirm attendance and enhance the accuracy of student attendance management. Through a visual interface, the number of people entering and leaving the classroom can be intuitively displayed, providing real-time and effective data support for school management. The use of morphological operation models to pre-process data helps improve image quality and enhance the accuracy of subsequent facial recognition and headcount monitoring. At the same time, it simplifies the subsequent model processing process, can remove small noise in the image or connect some discontinuous parts, thereby more accurately identifying the target outline, and combines dilation and erosion to achieve more complex morphological transformations. Using a combination of opening and closing operations, better image processing effects are achieved.

[0106] One embodiment of the present invention provides a real-time crowd flow statistics system based on Internet of Things information collection, wherein the display module includes:

[0107] The corresponding time attendance rate display module is used to store the attendance information of classrooms with classes after the local data center processes it according to the attendance rate of the corresponding class hours of the course, and display the attendance rate of each classroom just processed through a visual interface;

[0108] Form a display data module, which is used by the local data center to process the number of people entering and leaving the vacant classrooms, and arrange them according to the display format of the electronic class board to form the overview class board display data and the class board display data for each floor of each building;

[0109] The transmission module displays the overview shift board display data on the overview shift board, and transmits the shift board display data of each floor of each building to the electronic shift board device of the corresponding floor through the local area network communication method.

[0110] The working principle of the above technical solution is as follows: the local data center processes the attendance information of classrooms with classes, calculates the attendance rate of the corresponding class hours, and stores it; displays the attendance rate of each classroom through a visual interface and provides it to relevant management personnel for viewing; processes the information of idle classrooms and compiles it into the display format of electronic class signs, forming overview class sign display data and class sign display data for each floor of each building; displays the overview class sign display data on the overview class sign, and then transmits the class sign display data for each floor of each building to the electronic class sign equipment on the corresponding floor through local area network communication;

[0111] Students can log in to the campus network to view the overview class information and obtain information about available classrooms.

[0112] The effects of the above technical solution are: through the processing of attendance information, school administrators can clearly understand the attendance rate of the corresponding courses in each classroom, providing data support for teaching management; the visual interface displays the attendance rate of the classroom, so that administrators can grasp the student attendance in a timely manner, providing a basis for subsequent teaching management and intervention; the vacant classroom information is arranged in the format of an electronic class board and transmitted to the electronic class board equipment on the corresponding floor, helping students to quickly obtain vacant classroom information; students can easily obtain overview class board information through the campus network, thereby quickly obtaining vacant classroom information, improving the efficiency of campus resource utilization; data transmission is completed through LAN communication, ensuring the security and immediacy of information.

[0113] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A real-time crowd flow statistics method based on Internet of Things information collection, characterized in that: The method comprises: Obtain classroom surveillance images through cameras; The electronic class board processes the classroom monitoring image, saves the processed result and uploads it to the local data center; The local data center generates real-time crowd statistics and displays them to teachers and students through a visual interface.

2. The method for real-time crowd flow statistics based on Internet of Things information collection according to claim 1 is characterized in that: Use cameras to obtain classroom monitoring images, including: When the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom, the electronic class sign corresponding to the classroom will query whether there is a class within the preset time period; If the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send control signals to the cameras inside the classroom and at the classroom door. The cameras inside the classroom and at the classroom door receive the control signals, start monitoring, and send the monitoring images to the corresponding electronic class board; If the electronic class board finds that the corresponding classroom has no classes within the preset time period, it uses the Zigbee communication protocol to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

3. The method for real-time crowd flow statistics based on Internet of Things information collection according to claim 1 is characterized in that: If the electronic class board finds that the corresponding classroom has no classes within the preset time period, it uses the Zigbee communication protocol to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring transmission frequency obtained by the frequency calculation model. The frequency calculation model is as follows: F = k1*[α*FD+(1-α)*(1-S)*Presence] Where F is the frequency, FD represents the difference between the previous and next frames of the surveillance video, Presence is a binary variable representing the presence of a person, which is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0, α is the weight used to balance the difference between the frames and the structural similarity index, k1 is a constant whose value range is [15,30] and is used to adjust the monitoring transmission frequency, and S is the structural similarity index. Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2 , where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes. I(x,y)=1-|L x -L y | / L max Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

4. The method for real-time crowd flow statistics based on Internet of Things information collection according to claim 1 is characterized in that: The electronic class board processes the classroom monitoring screen, saves the processed results and uploads them to the local data center, including: When the electronic class card receives surveillance footage from the classroom and at the classroom door, it first compares the faces captured by the surveillance footage at the classroom door with the faces of the class that will start the class stored in the electronic class card. It then continuously uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of people. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device, who then further confirms the attendance. The teacher's confirmation information is then uploaded to the local data center. When the electronic class card receives surveillance footage of the classroom door, it first pre-processes the data, then uses the YOLOv5s model to obtain the number of people entering and leaving the classroom, displays it through a visual interface, and then uploads it to the local data center. During data preprocessing, the following models can be used to optimize it: Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

5. The method for real-time crowd flow statistics based on Internet of Things information collection according to claim 1 is characterized in that: The local data center generates real-time crowd statistics and displays them to teachers and students through a visual interface, including: After processing the attendance information of classrooms with classes, the local data center stores it according to the attendance rate of the corresponding class hours of the course, and displays the attendance rate of each classroom just processed through a visual interface; The local data center processes the number of people entering and leaving the vacant classrooms and arranges them according to the display format of the electronic class board, forming the overview class board display data and the class board display data for each floor of each building; The overview class board display data is displayed on the overview class board, and the class board display data of each floor of each building is transmitted to the electronic class board device on the corresponding floor through the local area network communication method. Students can view the overview class board information by logging into the campus network to obtain the available classroom information.

6. A real-time crowd statistics system based on Internet of Things information collection, characterized by: The system comprises: Monitoring module, used to obtain classroom monitoring images through cameras; A processing module, used for the electronic class board to process the classroom monitoring screen, save the processed results and upload them to the local data center; The display module is used for the local data center to generate real-time crowd statistics results and display them to the user through a visual interface. Teachers and students.

7. The real-time crowd flow statistics system based on Internet of Things information collection according to claim 6 is characterized in that: The monitoring module includes: The module for detecting someone entering the classroom is used to check whether there is a class in the preset time period on the electronic class board corresponding to the classroom when the infrared sensor in the classroom detects that there is no one in the classroom and the infrared sensor at the classroom door detects that someone has entered the classroom; Start the camera module inside the classroom and at the classroom door. When the electronic class board finds that the corresponding classroom has classes within the preset time period, it uses the Zigbee communication protocol to send a control signal to the camera inside the classroom and at the classroom door. The camera inside the classroom and at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board; Start the camera module at the classroom door. When the electronic class board finds that the corresponding classroom has no classes within the preset time period, the Zigbee communication protocol is used to send a control signal to the camera at the classroom door. The camera at the classroom door receives the control signal, starts monitoring, and sends the monitoring image to the corresponding electronic class board according to the monitoring frequency obtained by the frequency calculation model.

8. The real-time crowd statistics system based on Internet of Things information collection according to claim 6 is characterized in that: The module for starting the camera at the classroom door includes a frequency calculation model module, F = k1*[α*FD+(1-α)*(1-S)*Presence] Where F is the frequency, FD represents the difference between the previous and next frames of the surveillance video, Presence is a binary variable representing the presence of a person, which is 1 when the infrared sensor in the classroom detects that someone has entered the classroom, otherwise it is 0, α is the weight used to balance the difference between the frames and the structural similarity index, and the value range of α is (0,1), k1 is a constant used to adjust the monitoring transmission frequency, and the value range of k1 is [15,30], S is the structural similarity index, Where x and y are the two images to be compared, μ1 and μ2 represent the means of the two images, σ1*σ1 represents the variance of image x, σ2*σ2 represents the variance of image y, and σ12 represents the covariance of the two images. C1 and C2 are constants used to avoid the case where the denominator is 0. The value of the constant varies depending on the color range. In general, C1 = (K1*L) 2 , C2=(K2*L) 2 , where L is the range of pixel values, L-1 is the maximum pixel value, K1 and K2 are constants less than 1, β is a trade-off factor used to balance the impact of structural similarity and light and shadow similarity on the overall similarity, and I(x,y) is a similarity function used to measure light and shadow changes. I(x,y)=1-|L x -L y | / L max Among them, L x and L y Represents the illumination intensity of the two images, L max is the maximum light intensity.

9. The real-time crowd flow statistics system based on Internet of Things information collection according to claim 6 is characterized in that: The processing module includes: The class-specific processing module is used to compare the faces captured by the surveillance footage from the classroom and at the classroom door with the faces of the class that will start the class, which is stored in the electronic class card. It then uses the YOLOv5s model to detect the number of heads in the classroom surveillance video to obtain the number of people in the class. If a face mismatch is found or the number of matched faces is inconsistent with the number of heads in the classroom, a message is sent to the corresponding teacher's electronic device for further confirmation of the teacher's attendance. The teacher's confirmation information is then uploaded to the local data center. The processing module for idle classrooms is used to pre-process the data when the electronic class card receives the surveillance image of the classroom door. Then, the YOLOv5s model is used to obtain the number of people entering and leaving the classroom. The data is displayed through a visual interface and then uploaded to the local data center. During the data pre-processing process, the following models can be used to optimize it. Among them, Θ(A,B) represents the new morphological operation, A represents the input image, and B represents the structural element, which is used to define the shape and size of the expansion and corrosion operations. Indicates that the image A is expanded to obtain the result after expansion. Express The result is eroded to obtain the final output image.

10. The real-time crowd flow statistics system based on Internet of Things information collection according to claim 6 is characterized in that: The display module includes: The corresponding time attendance rate display module is used to store the attendance information of classrooms with classes after the local data center processes it according to the attendance rate of the corresponding class hours of the course, and display the attendance rate of each classroom just processed through a visual interface; Form a display data module, which is used by the local data center to process the number of people entering and leaving the vacant classrooms, and arrange them according to the display format of the electronic class board to form the overview class board display data and the class board display data for each floor of each building; The transmission module displays the overview class board display data on the overview class board, and transmits the class board display data of each floor of each building to the electronic class board equipment on the corresponding floor through the local area network communication method. Students can view the overview class board information by logging into the campus network to obtain the available classroom information.

Images