Intelligent operation safety management and control method and system based on personnel behavior analysis and position tracking

Abstract

The invention discloses an intelligent operation safety management and control method and system based on personnel behavior analysis and position tracking, and the method comprises the steps: building a three-dimensional coordinate system of an operation region, determining the position of a camera device and the entrance and exit position of the operation region in the coordinate system, and carrying out the positioning of the camera device; initializing a shooting angle and a scanning three-dimensional area range of the camera equipment according to the position information; identifying personnel, calibrating coordinates of the personnel in the three-dimensional coordinate system, and controlling corresponding camera equipment to track the personnel; setting a risk management and control area, and judging whether border crossing behaviors exist or not according to the personnel position information; a visual algorithm based on deep learning is adopted to perform real-time analysis on personnel behaviors and judge whether illegal behaviors exist or not. The visual technology is adopted for personnel behavior analysis and position tracking, personnel positions are accurately tracked, the visual algorithm based on deep learning is combined, personnel behaviors are analyzed in real time, whether illegal behaviors exist or not is judged, and intelligent operation safety management and control are achieved.

Description

Technical Field

[0001] The present invention relates to the field of computer vision, and in particular to an intelligent operation safety management and control method and system based on personnel behavior analysis and position tracking. Background Art

[0002] The essence of operation safety control is the management and control of personnel behavior. At present, the implementation of operation safety control is generally through general and targeted rules and regulations, personnel monitoring, video surveillance, etc. This method will incur a large labor cost, the level of intelligence is low, and the control effect is heavily dependent on the working status of personnel. Personnel monitoring has the following shortcomings: poor execution, such as the supervisor does not strictly follow the operating procedures, such as not checking and implementing safety measures item by item against the requirements of the license, or not confirming the on-site safety measures, emergency plans, etc.; lack of responsibility: the supervisor leaves the supervisory post without authorization, or conducts multi-point monitoring, resulting in inadequate supervision. Summary of the invention

[0003] Purpose of the invention: The present invention aims to provide an intelligent operation safety management method based on personnel behavior analysis and location tracking, which reduces labor costs and increases the efficiency of operation safety management; another purpose of the present invention is to provide an intelligent operation safety management system based on personnel behavior analysis and location tracking.

[0004] Technical solution: The intelligent operation safety management method based on personnel behavior analysis and location tracking of the present invention comprises the following steps:

[0005] S1: Construct the three-dimensional coordinate system of the working area;

[0006] S2: Determine the position of the camera device in the coordinate system according to the position of the camera device in the work area;

[0007] S3: Determine the entrance and exit positions of the work area, and initialize the camera shooting angle and scanning three-dimensional area range according to the position information;

[0008] S4: Identify the person and calibrate the coordinates of the person in the three-dimensional coordinate system, and control the corresponding camera device to track the person according to the calibrated coordinates;

[0009] S5: Set up risk control areas and determine whether there is any cross-border behavior based on personnel location information;

[0010] S6: Use deep learning-based visual algorithms to analyze people’s behavior in real time to determine whether there are any violations;

[0011] S7: Send the violation of security control to the system, retain video and picture records, and support record download.

[0012] Furthermore, step S1 is specifically as follows:

[0013] S11: Determine the scope of the operation area and confirm the area of ​​the operation according to the actual situation on site;

[0014] S12: Determine the origin position of the three-dimensional coordinate system according to the principle of positive coordinates.

[0015] Furthermore, step S2 is specifically as follows:

[0016] Camera equipment includes high-definition ball cameras and high-definition gun cameras;

[0017] S21: Obtaining the spatial distribution of the working area and the observation range information of the camera equipment;

[0018] S22: Calculate the minimum number of camera devices that can cover the entire operation area based on the information obtained in S21;

[0019] S23: Calculate the spacing between the cameras according to the number of the cameras, and calibrate the positions of the cameras in the three-dimensional coordinate system.

[0020] Furthermore, step S3 is specifically as follows:

[0021] S31: Determine the entrance and exit locations of the operation area and set up a high-definition gun camera to observe the entrance and exit;

[0022] S32: Determine the shooting angle of the camera device during initialization according to the camera device position information determined in S2 and the entrance and exit positions determined in S31;

[0023] S33: Initialize the shooting position of the camera device when there is no one in the working area.

[0024] Further, step S4 is specifically as follows:

[0025] S41: Recognize human body contour;

[0026] S42: Calculating the three-dimensional coordinates of the human body according to the maximum circumscribed rectangle of the recognized human body contour and the corresponding camera device position;

[0027] S43: Connecting the camera device to the streaming media service, calling the camera device at the corresponding position according to the three-dimensional coordinates of the human body to adjust the angle so that the human body is always at the center of the video screen;

[0028] S44: when the coordinates of the human body move out of the shooting range of the camera device, the corresponding camera device is calculated and called according to the three-dimensional coordinates of the human body, the position coordinates of the camera device and the shooting range;

[0029] S45: Repeat steps S43 and S44 to implement personnel tracking.

[0030] Furthermore, step S5 is specifically as follows:

[0031] S51: Determine the risk control area in the three-dimensional coordinate system;

[0032] S52: judging whether the risk control area and the three-dimensional coordinates of the human body have overlapping parts according to the three-dimensional coordinates of the human body;

[0033] S53: Calculate the overlap ratio, and if it is greater than a threshold, determine that there is an out-of-bounds behavior.

[0034] Furthermore, step S6 is specifically as follows:

[0035] S61: Using a streaming media service to obtain a real-time video stream of a corresponding camera device;

[0036] S62: Setting a real-time video stream sampling period to obtain a picture to be identified;

[0037] S63: According to the configuration of corresponding recognition models and analysis programs for different illegal behaviors, human behavior analysis is performed on the image to be recognized obtained in S62;

[0038] S64: Determine whether there is any violation based on the results of personnel behavior analysis.

[0039] Further, step S63 is specifically as follows:

[0040] S631: Determine the deep learning-based visual algorithm based on the actual safety control needs on site; the actual safety control needs on site include the recognition of personnel falls, single-person operations, wearing of safety helmets, wearing of work clothes, smoking and fireworks;

[0041] S632: Use the data set to train the algorithm architecture of the visual algorithm, analyze the images of human behavior, and evaluate the results;

[0042] S633: Determine whether to resample and optimize the training according to the evaluation result;

[0043] S634: Repeat steps S632 and S633 to obtain a personnel behavior analysis model, analyze personnel behavior, and determine whether there is any violation.

[0044] Further, step S7 is specifically as follows:

[0045] S71: Build intelligent operation safety management and control hardware and software systems;

[0046] S72: The violation determined in step 6 is sent to the system, the video and pictures of the violation are stored, and the record download is supported.

[0047] The intelligent operation safety management and control system based on personnel behavior analysis and position tracking described in the present invention includes: a camera device for collecting personnel behavior; a streaming media server for obtaining video streams and controlling the camera device; a video analysis server for training or reasoning analysis of the collected images; and a system server for executing the steps of the above-mentioned method to realize the execution logic of the system.

[0048] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. The present invention adopts visual technology to perform personnel behavior analysis and position tracking, and combines it with a visual algorithm based on deep learning to perform real-time analysis of personnel behavior, determine whether there are any violations, and realize intelligent operation safety management and control; 2. The present invention can accurately track the position of personnel, analyze and identify personnel operation behaviors, record violations, select corresponding algorithm architectures for different personnel behavior scenarios, and identify violations. It has the characteristics of high precision, high efficiency, and accurate recognition. It also supports model optimization training and enhances system robustness, which can effectively improve the efficiency of operation safety management and control and reduce labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the intelligent operation safety management method based on personnel behavior analysis and location tracking of the present invention;

[0050] Figure 2 This is a top view of the working area;

[0051] Figure 3 This is a schematic diagram of the installation location of the camera equipment in the working area;

[0052] Figure 4 It is a schematic diagram of the personnel location and the area that has crossed the boundary;

[0053] Figure 5 This is a software functional architecture diagram of the intelligent operation safety management and control system based on personnel behavior analysis and location tracking of the present invention;

[0054] Figure 6 This is the hardware architecture diagram of the intelligent operation safety management system based on personnel behavior analysis and location tracking of the present invention DETAILED DESCRIPTION

[0055] The present invention will be further described below in conjunction with the accompanying drawings.

[0056] The intelligent operation safety management and control method based on personnel behavior analysis and location tracking of the present invention comprises the following steps:

[0057] S1: Establish a three-dimensional coordinate system according to the working area.

[0058] S11: Figure 2As shown, determine the scope of the operation area and confirm the area where the operation may occur based on the actual situation on site;

[0059] S12: Determine the origin position of the three-dimensional coordinate system according to the principle of positive coordinates.

[0060] The "positive coordinate principle" mentioned here means that the three coordinate axes (X, Y, Z) in three-dimensional space are only positive.

[0061] S2: Determine the position of the indoor camera equipment in the working area, and determine the position of the camera equipment in the coordinate system.

[0062] The specific steps include:

[0063] S21: Obtain the spatial distribution of the operation area according to the scene business needs, and determine the observation range information of the camera equipment according to the model and performance of the camera equipment;

[0064] S22: Based on the information obtained in S21, the minimum number of cameras that can cover the entire operation area is calculated. The calculation principles are mainly two points. First, the camera equipment must fully cover the operation area without any blind spots. Second, the number of camera equipment should be saved as much as possible to reduce costs. Figure 3 As shown;

[0065] S23: Calculate the spacing between the cameras according to the number of the cameras in S22, and calibrate the positions of the cameras in the three-dimensional coordinate system.

[0066] S3: Determine the entrance and exit positions of the work area, and initialize the shooting angle of the camera device and the scanning three-dimensional area range according to the position information.

[0067] S31: Confirm the entrance and exit locations of the operation area based on the actual situation on site, and set up a high-definition gun camera to observe the entrance and exit. The gun camera is mainly responsible for using artificial intelligence technology to count the number of people entering the area;

[0068] S32: Determine the shooting angle of the camera device during initialization according to the camera device position information determined in S2 and the entrance and exit positions determined in S31. The principle of determining the initial shooting angle is that the camera device can completely capture the person when the person enters the working area and the person is in the center of the picture as much as possible.

[0069] S4: Identify the person, calibrate him / her in a three-dimensional coordinate system, and control the corresponding camera device to track him / her according to the coordinates.

[0070] The specific steps include:

[0071] S41: Recognize human body contours through artificial intelligence technology;

[0072] S42: Calculating the three-dimensional coordinates of the human body according to the maximum circumscribed rectangle of the human body contour identified in S41 and the corresponding camera device position, wherein after obtaining the maximum circumscribed rectangle, the proportion of the human body in the screen is calculated, and the three-dimensional coordinates of the human body are calculated in combination with the current angle of the camera;

[0073] S43: Connect the camera device to the streaming media service, and call the camera device at the corresponding position according to the three-dimensional coordinates of the human body in S42 to adjust the angle so that the human body is always in the center of the video screen. The system designs a unified interface streaming media service based on the SDK of the camera devices of various manufacturers. As the human body moves and gradually moves away from the center area of ​​the camera device, the corresponding camera device angle is moved according to the calculated three-dimensional coordinates of the human body so that the human body is always in the center of the video screen;

[0074] S44: when the coordinates of the human body move out of the shooting range of the camera device, the corresponding camera device is calculated and called according to the three-dimensional coordinates of the human body, the position coordinates of the camera device and the shooting range in S42, wherein as the human body continues to move, it will eventually move out of the monitoring range of the current camera device, and according to the three-dimensional coordinates of the human body and the known position and shooting range of the camera device, the corresponding camera device is called by using the streaming media service to continue monitoring the moving person;

[0075] S45: Repeat steps S43 and S44 to implement personnel tracking.

[0076] S5: Draw the risk control area and determine whether there is any cross-border behavior based on the personnel location information.

[0077] S51: Figure 4 As shown, the risk control area is delineated in the three-dimensional coordinate system;

[0078] S52: judging whether the risk control area and the three-dimensional coordinates of the human body overlap with each other according to the three-dimensional coordinates of the human body determined in S4, wherein the coordinates (X, Y, Z) need to be judged respectively;

[0079] S53: Calculate the overlap ratio according to S52, and if it is greater than a threshold, determine that there is an out-of-bounds behavior, wherein the ratio of the risk control area to the human body coordinate area is calculated based on the coordinate values ​​of the overlapping part.

[0080] S6: Use deep learning algorithms in the field of computer vision to analyze people’s behavior in real time to determine whether there are any violations.

[0081] S61: From S4, it can be known that the operator is always at the center of the camera equipment, and uses the streaming media service to obtain the real-time video stream of the corresponding camera equipment. The system supports the collection and processing of RTSP video streams of different manufacturers through the streaming media service;

[0082] S62: Setting a real-time video stream sampling period to obtain a picture to be recognized, wherein the video stream sampling period is configured according to the system computing power, and the sampling period directly affects the computing resources required for recognition;

[0083] S63: Configure different recognition models and analysis programs according to different violations, and analyze the behavior of people in the pictures collected by S62. According to the actual safety control needs on site, such as people falling, single-person operation, wearing of safety helmets, wearing of work clothes, smoking, fireworks recognition, etc., according to the above needs, screen the needs that can be solved by image analysis, select the appropriate computer vision algorithm architecture, and use the corresponding pre-trained model for different algorithm architectures to infer the image of people's behavior and evaluate the results. If the evaluation result is lower than the preset threshold, it is necessary to optimize the training for this scene. In order to improve the recognition accuracy, the system can be configured to repeatedly evaluate and optimize to guide the evaluation results to meet expectations;

[0084] S64: Based on the results of the personnel behavior analysis in S63, determine whether there is any violation, wherein the system can issue graded alarms based on different violation types.

[0085] S7: Send the violation of security control to the system, retain video and picture records, and support record download.

[0086] The specific steps include:

[0087] S71: judging the personnel behavior analysis and recognition results obtained in S6, wherein the recognition results that comply with the operation safety regulations are ignored;

[0088] S72: Build intelligent operation safety management software (such as Figure 5 shown) and hardware (as shown Figure 6 The system includes: a camera device for collecting personnel behavior; a streaming media server for obtaining video streams and controlling the camera device; a video analysis server for training or reasoning analysis of the collected images; and a system server for implementing the execution logic of the entire system.

[0089] S73: Send the violation of security control in S71 to the system, where the system message bus is responsible for receiving the sent information;

[0090] S74: Supports the data storage of S73 uploaded to the system, and saves the video and pictures when it occurs. It also supports record downloading. The uploaded data is taken out from the system message bus according to the data type and stored in relational and non-relational databases for persistence, and supports conditional query and downloading.

Claims

1. An intelligent operation safety management and control method based on personnel behavior analysis and location tracking, characterized in that: The steps include: S1: Construct the three-dimensional coordinate system of the working area; S2: Determine the position of the camera device in the coordinate system according to the position of the camera device in the work area; S3: Determine the entrance and exit positions of the work area, and initialize the camera shooting angle and scanning three-dimensional area range according to the position information; S4: Identify the person and calibrate the coordinates of the person in the three-dimensional coordinate system, and control the corresponding camera device to track the person according to the calibrated coordinates; S5: Set up risk control areas and determine whether there is any cross-border behavior based on personnel location information; S6: Use deep learning-based visual algorithms to analyze people’s behavior in real time to determine whether there are any violations; S7: Send the violation of security control to the system, retain video and picture records, and support record download.

2. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 1 is characterized in that: Step S1 is specifically as follows: S11: Determine the scope of the operation area and confirm the area of ​​the operation according to the actual situation on site; S12: Determine the origin position of the three-dimensional coordinate system according to the principle of positive coordinates.

3. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 2 is characterized in that: Step S2 is specifically as follows: Camera equipment includes high-definition ball cameras and high-definition gun cameras; S21: Obtaining the spatial distribution of the working area and the observation range information of the camera equipment; S22: Calculate the minimum number of camera devices that can cover the entire operation area based on the information obtained in S21; S23: Calculate the spacing between the cameras according to the number of the cameras, and calibrate the positions of the cameras in the three-dimensional coordinate system.

4. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 3 is characterized in that: Step S3 is as follows: S31: Determine the entrance and exit locations of the operation area and set up a high-definition gun camera to observe the entrance and exit; S32: Determine the shooting angle of the camera device during initialization according to the camera device position information determined in S2 and the entrance and exit positions determined in S31; S33: Initialize the shooting position of the camera device when there is no one in the working area.

5. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 4 is characterized in that: Step S4 is specifically as follows: S41: Recognize human body contour; S42: Calculating the three-dimensional coordinates of the human body according to the maximum circumscribed rectangle of the recognized human body contour and the corresponding camera device position; S43: Connecting the camera device to the streaming media service, calling the camera device at the corresponding position according to the three-dimensional coordinates of the human body to adjust the angle so that the human body is always at the center of the video screen; S44: when the coordinates of the human body move out of the shooting range of the camera device, the corresponding camera device is calculated and called according to the three-dimensional coordinates of the human body, the position coordinates of the camera device and the shooting range; S45: Repeat steps S43 and S44 to implement personnel tracking.

6. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 5 is characterized in that: Step S5 is specifically as follows: S51: Determine the risk control area in the three-dimensional coordinate system; S52: judging whether the risk control area and the three-dimensional coordinates of the human body have overlapping parts according to the three-dimensional coordinates of the human body; S53: Calculate the overlap ratio, and if it is greater than a threshold, determine that there is an out-of-bounds behavior.

7. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 6 is characterized in that: Step S6 is specifically as follows: S61: Using a streaming media service to obtain a real-time video stream of a corresponding camera device; S62: Setting a real-time video stream sampling period to obtain a picture to be identified; S63: According to the configuration of corresponding recognition models and analysis programs for different illegal behaviors, human behavior analysis is performed on the image to be recognized obtained in S62; S64: Determine whether there is any violation based on the results of personnel behavior analysis.

8. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 7 is characterized in that: Step S63 is specifically as follows: S631: Determine the deep learning-based visual algorithm based on the actual safety control needs on site; the actual safety control needs on site include the recognition of personnel falls, single-person operations, wearing of safety helmets, wearing of work clothes, smoking and fireworks; S632: Use the data set to train the algorithm architecture of the visual algorithm, analyze the images of human behavior, and evaluate the results; S633: Determine whether to resample and optimize the training according to the evaluation result; S634: Repeat steps S632 and S633 to obtain a personnel behavior analysis model, analyze personnel behavior, and determine whether there is any violation.

9. The intelligent operation safety management and control method based on personnel behavior analysis and location tracking according to claim 8 is characterized in that: Step S7 is specifically as follows: S71: Build intelligent operation safety management and control hardware and software systems; S72: The violation determined in step 6 is sent to the system, the video and pictures of the violation are stored, and the record download is supported.

10. An intelligent operation safety management and control system based on personnel behavior analysis and location tracking, characterized in that: include: Camera equipment, used to collect information about people's behavior; Streaming media server, used to obtain video streams and control camera equipment; A video analysis server is used to perform training or reasoning analysis on the collected images; a system server is used to execute the steps of the method described in any one of claims 1 to 7 and implement the execution logic of the system.

Images