Construction site safety monitoring method and system based on deep learning
By deploying high-definition cameras at construction sites, using deep learning technology for image processing and risk assessment, and building climbing complexity, safety belt wear, personnel risk and environmental risk assessment models, the problems of existing technologies such as the difficulty in achieving real-time monitoring of the entire area and incomplete safety risk assessment have been solved, and intelligent safety management of construction sites has been achieved.
Patent Information
- Application Number
- CN202511154754.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-18
- Publication Date
- 2025-09-16
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing construction safety monitoring methods rely on manual inspections and experience-based judgments, making it difficult to achieve real-time monitoring of the entire area. They are unable to effectively assess the dynamic data of workers and the wear risk of safety belts, resulting in incomplete safety risk assessments.
Using a deep learning-based method, high-definition cameras collect images of workers, perform preprocessing and skeleton recognition, and build climbing complexity, safety belt wear, personnel risk and environmental risk assessment models to comprehensively evaluate the safety of the work scene and issue early warnings.
It has achieved a comprehensive dynamic risk assessment and early warning of construction site workers, environment and equipment, improved the level of intelligent safety management at the construction site, and ensured the safety and real-time performance of operations.
Smart Images
Figure CN120655110A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of construction safety monitoring, specifically a construction site safety monitoring method and system based on deep learning. Background Art
[0002] At construction sites, especially in environments involving height operations, safety accidents such as falls from heights are very likely to occur due to the high working height, complex structure, and frequent personnel activities. Traditional safety monitoring methods mainly rely on manual inspections, safety officers' experience judgment, or regular inspections. Manual inspections are limited by manpower and cannot achieve real-time monitoring of the entire area. It is difficult to identify high-risk operations in a timely manner, and it is impossible to evaluate safety risks based on dynamic data such as the posture, behavior, and tool use of the workers. At the same time, it is impossible to assess the risk of safety belt wear during use, and the assessment of personnel work safety is not comprehensive.
[0003] This application realizes comprehensive dynamic risk assessment and early warning of construction site workers, environment and equipment through image acquisition, posture recognition, seat belt wear assessment, personnel and environmental risk assessment, and improves the level of intelligent safety management on construction sites. Summary of the Invention
[0004] In response to the shortcomings of the existing technology, this application proposes a construction site safety monitoring method and system based on deep learning.
[0005] To achieve the above objectives, this application provides the following technical solutions: The construction site safety monitoring method and system based on deep learning includes the following specific steps: Acquire images of personnel working at construction sites, pre-process the images to identify personnel working conditions, and simultaneously obtain working environment parameters; Construct a climbing complexity assessment model and import the collected personnel work conditions into the climbing complexity assessment model to evaluate the climbing complexity of the work area; Construct a seat belt wear assessment model, import seat belt usage records into the seat belt wear assessment model to evaluate the wear of seat belts during use and assess the safety of seat belts; Build a personnel risk assessment model and import the basic information of the operators into the personnel risk assessment model to assess the risk situation of the operators themselves; Construct an environmental risk assessment model and import the environmental parameters of the operation period into the environmental risk assessment model to evaluate the operational risks caused by the environment; Construct a comprehensive risk assessment model, import climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the operation scenario and issue early warning.
[0006] Preferably, the steps of acquiring images of personnel working at a construction site, pre-processing the images to identify personnel working conditions, and acquiring working environment parameters include the following specific steps: S11. Using multiple high-definition cameras deployed in specific areas of the construction site, real-time images of workers climbing and moving on the structure are collected, and the collected images are pre-processed by frame separation, denoising, and enhancement. S12, identifying the location frame of the operator's area, cropping the entire frame image into a target area image for subsequent skeleton recognition, and using a mainstream human posture estimation algorithm to identify skeleton points of the person in the image frame to obtain a skeleton point set; S13. Obtain basic information of workers and environmental parameters of working hours and time periods.
[0007] Preferably, the step of constructing a climbing complexity assessment model and importing the collected personnel work conditions into the climbing complexity assessment model to assess the climbing complexity of the work area includes the following specific steps: S21. Obtain a set of trajectories of the operator's body center point in consecutive frames, and calculate the degree of trajectory tortuosity using the curvature variation coefficient. The curvature variation coefficient is the ratio of the total length of the trajectory's broken line segments to the straight-line distance between the starting and ending points. The curvature variation coefficient tends to increase with more curved paths. The curvature variation coefficient is used to reflect the increased difficulty of non-straight climbing. The height fluctuation index is obtained by calculating the standard deviation of the height of the operator's body center point per unit time. The height fluctuation index reflects the intensity and frequency of the operator's vertical movements and is used to determine climbing rhythm and altitude stability. S22. Calculate the climbing speed for each frame based on the displacement vectors of the skeleton center point in different frames to obtain a climbing speed sequence. Substitute the climbing speed into the speed fluctuation standard deviation calculation formula to evaluate the operator's climbing speed fluctuation by calculating the speed standard deviation. The greater the speed fluctuation, the more unstable the climbing rhythm, and the higher the potential risk of falling and imbalance. S23. Segmenting each image frame into background and obstacles using a semantic segmentation model to identify occluding and interfering structures on the climbing path. The obstacle interference is assessed using an obstacle occlusion ratio, where the obstacle occlusion ratio is the ratio of the intersection of the obstacle area in the image and the climbing path area predicted by the skeleton to the climbing path area predicted by the skeleton. A larger obstacle occlusion ratio indicates more severe path interference. S24. Substitute the curvature variation coefficient, height fluctuation index, speed fluctuation standard deviation, and obstacle shielding ratio into the climbing complexity calculation formula to evaluate the complexity of the climbing behavior of the person in a specific time period. The climbing complexity calculation formula is: , where the curvature variation coefficient is F, the height fluctuation index is H, the speed fluctuation standard deviation is V, and the obstacle occlusion ratio is D. 、 、 and is the weight.
[0008] Preferably, the construction of the seat belt wear assessment model, importing the seat belt usage records into the seat belt wear assessment model to assess the wear of the seat belt during use, and assessing the safety of the seat belt include the following specific steps: S31. Obtain seat belt usage records, and substitute seat belt usage parameters into a seat belt wear index calculation formula to evaluate seat belt wear. The seat belt wear index calculation formula is: ,in, is the length of time the seat belt has been used, The maximum usage time of the seat belt. is the number of times the seat belt has been used, is the maximum number of times the seat belt is used, is the average task intensity level, is the maximum value of the task intensity level, is the time since the last maintenance, For maintenance cycle, 、 、 and is the weight, ; S32. Predict the safety belt wear index during use based on a deep learning neural network, obtain the roughness and hardness of the building surface, and at the same time obtain the contact area between the safety belt and the building surface, the safety belt wear index at the starting time, the operation time, the weight of the person and the climbing complexity, and construct a deep learning neural network whose input is the roughness and hardness of the building surface, the contact area between the safety belt and the building surface, the safety belt wear index at the starting time, the operation time, the weight of the person and the climbing complexity, and the output is the safety belt wear index during use. The specific steps are: obtain the roughness and hardness of the building surface during the historical use of the safety belt, and at the same time obtain the contact area between the safety belt and the building surface The contact surface of the building surface, the safety belt wear index at the starting time, the operation time, the weight and climbing complexity of the person, and the safety belt wear index before and after use are obtained. The historical data is divided into an 85% weighted and biased training set and a 15% weighted and biased test set. The 85% weighted and biased training set is input into the deep learning neural network model for training to obtain an initial deep learning neural network model. The initial deep learning neural network model is tested using the 15% weighted and biased test set, and the initial deep learning neural network model output that meets the maximum preset safety belt wear index accuracy during use is used as the deep learning neural network model. S33. Compare the seat belt wear index during use obtained by the neural network prediction with the seat belt threshold. If it is within the threshold range, the seat belt can continue to be used. If it exceeds the threshold range, replace the seat belt immediately and recalculate the seat belt wear index.
[0009] Preferably, the construction of the personnel risk assessment model and the importation of the basic information of the operating personnel into the personnel risk assessment model to assess the risk situation of the personnel themselves include the following specific steps: S41. Substitute the basic information parameters of the operator into the personnel risk calculation formula to evaluate the risk situation of the operator, wherein the personnel risk calculation formula is: ,in, The seat belt influence coefficient, P' is the seat belt wear index during use predicted by the neural network, P" is the seat belt threshold, and L is the working time of the operator since the start of work. is the shift length of the operator, BMI is the body mass index, 22 is the optimal BMI index, 13 is the maximum allowable body mass index deviation, E is the operator's age, and C is the operator's climbing experience level.
[0010] Preferably, the construction of the environmental risk assessment model and the importation of the environmental parameters of the operation period into the environmental risk assessment model to assess the operation risk caused by the environment include the following specific steps: Substitute the environmental parameters of the operation period into the environmental risk assessment formula to evaluate the environmental conditions during the operation. The environmental risk assessment formula is: ,in, is the i-th environmental risk factor, is the weight, and n is the number of environmental risk factors, where environmental risk factors include temperature, light intensity, wind speed, etc. The environmental risk factor calculation formula is: ,in, is the i-th environmental parameter of the operation period, is the suitable value of the i-th environmental parameter.
[0011] Preferably, the construction of a comprehensive risk assessment model, importing climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the operation scene, and issuing an early warning includes the following specific steps: Substitute climbing complexity, personnel risk, and environmental risk into the comprehensive safety score formula for the operation scenario and obtain the comprehensive safety score for the operation scenario through weighted addition. Compare the comprehensive safety score for the operation scenario with the safety score threshold, set the warning level, and perform normal operations for the safety level. Stop operations for non-safety levels.
[0012] The deep learning-based construction site safety monitoring system is implemented based on the above-mentioned deep learning-based construction site safety monitoring method, and specifically includes: The data acquisition module is used to obtain images of personnel working at the construction site, pre-process the images to identify the personnel working conditions, and obtain working environment parameters; The climbing complexity assessment module is used to assess the climbing complexity of the work area based on the collected personnel work conditions; The seat belt wear assessment module is used to evaluate the wear of seat belts during use and the safety of seat belts through seat belt usage records; The personnel risk assessment module is used to assess the risk situation of personnel themselves based on their basic information; Environmental risk assessment module, used to evaluate the operational risks caused by the environment through environmental parameters during the operation period; The comprehensive risk assessment module is used to evaluate the overall safety situation of the operation scenario through climbing complexity, safety belt wear index, personnel risk and environmental risk, and provide early warning.
[0013] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor; The processor executes the above-mentioned deep learning-based construction site safety monitoring method by calling the computer program stored in the memory.
[0014] A computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned construction site safety monitoring method based on deep learning.
[0015] Compared with the prior art, the present invention has the following advantages: The present application obtains images of personnel working at a construction site, pre-processes the images to identify the personnel working conditions, obtains working environment parameters at the same time, constructs a climbing complexity assessment model, imports the collected personnel working conditions into the climbing complexity assessment model to evaluate the climbing complexity of the working area, constructs a safety belt wear assessment model, imports the safety belt usage records into the safety belt wear assessment model to evaluate the wear of the safety belt during use, evaluates the safety of the safety belt, constructs a personnel risk assessment model, imports the basic information of the operating personnel into the personnel risk assessment model to evaluate the personnel's own risk situation, constructs an environmental risk assessment model, imports the environmental parameters of the working period into the environmental risk assessment model to evaluate the working risk situation caused by the environment, constructs a comprehensive risk assessment model, imports the climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the working scene and issue early warnings, thereby improving the level of intelligent safety management at the construction site. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 This is a schematic diagram of the overall process of the construction site safety monitoring method based on deep learning in this application; Figure 2 This is a flowchart of the seat belt wear index assessment process during the use of this application; Figure 3 A comprehensive safety scoring and assessment flow chart for the operation scenario; Figure 4 This is a schematic diagram of the overall framework of the construction site safety monitoring system based on deep learning in this application. DETAILED DESCRIPTION
[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.
[0018] Example 1 See also Figures 1-4 , this application provides an embodiment: Figure 1 As shown in FIG, a construction site safety monitoring method based on deep learning includes the following specific steps: Acquire images of personnel working at construction sites, pre-process the images to identify personnel working conditions, and simultaneously obtain working environment parameters; Construct a climbing complexity assessment model and import the collected personnel work conditions into the climbing complexity assessment model to evaluate the climbing complexity of the work area; Construct a seat belt wear assessment model, import seat belt usage records into the seat belt wear assessment model to evaluate the wear of seat belts during use and assess the safety of seat belts; Build a personnel risk assessment model and import the basic information of the operators into the personnel risk assessment model to assess the risk situation of the operators themselves; Construct an environmental risk assessment model and import the environmental parameters of the operation period into the environmental risk assessment model to evaluate the operational risks caused by the environment; Construct a comprehensive risk assessment model, import climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the operation scenario and issue early warning.
[0019] In this embodiment, it should be specifically explained that obtaining images of personnel working at a construction site, pre-processing the images to identify personnel working conditions, and obtaining working environment parameters include the following specific steps: S11. Utilize multiple high-definition cameras deployed in specific areas of the construction site to capture real-time images of workers climbing and moving on structures, such as scaffolding and rebar racks. Perform frame separation, denoising, and enhancement preprocessing on the captured images, including target detection, background separation, lighting enhancement, and image scaling. Centrally access and manage all cameras to ensure the continuity and time synchronization of the captured video streams. S12. Identify the location frame of the operator's area, crop the entire frame image into the target area image for subsequent skeleton recognition, and use the mainstream human pose estimation algorithm OpenPose to identify the skeleton points of the person in the image frame to obtain a skeleton point set; S13. Obtain basic information of workers and environmental parameters of working hours and time periods.
[0020] In this embodiment, it should be specifically explained that building a climbing complexity assessment model and importing the collected personnel work conditions into the climbing complexity assessment model to assess the climbing complexity of the work area includes the following specific steps: S21. Obtain a set of trajectories of the operator's body center point in consecutive frames, and calculate the degree of trajectory tortuosity using the curvature variation coefficient. The curvature variation coefficient is the ratio of the total length of the trajectory's broken line segments to the straight-line distance between the starting and ending points. The curvature variation coefficient tends to increase with more curved paths. The curvature variation coefficient is used to reflect the increased difficulty of non-straight climbing. The height fluctuation index is obtained by calculating the standard deviation of the height of the operator's body center point per unit time. The height fluctuation index reflects the intensity and frequency of the operator's vertical movements and is used to determine climbing rhythm and altitude stability. S22. Calculate the climbing speed for each frame based on the displacement vectors of the skeleton center point in different frames to obtain a climbing speed sequence. Substitute the climbing speed into the speed fluctuation standard deviation calculation formula to evaluate the operator's climbing speed fluctuation by calculating the speed standard deviation. The greater the speed fluctuation, the more unstable the climbing rhythm, and the higher the potential risk of falling and imbalance. S23. Use the semantic segmentation model DeepLabv3+ to segment the background and obstacles of each frame image, identify the occluding and interfering structures on the climbing path, and evaluate the interference of obstacles using the obstacle occlusion ratio. The obstacle occlusion ratio is the ratio of the intersection of the obstacle area in the image and the climbing path area predicted by the skeleton to the climbing path area predicted by the skeleton. A larger obstacle occlusion ratio indicates more severe path interference. S24. Substitute the curvature variation coefficient, height fluctuation index, speed fluctuation standard deviation, and obstacle shielding ratio into the climbing complexity calculation formula to evaluate the complexity of the climbing behavior of the person in a specific time period. The climbing complexity calculation formula is: , where the curvature variation coefficient is F, the height fluctuation index is H, the speed fluctuation standard deviation is V, and the obstacle occlusion ratio is D. 、 、 and is the weight.
[0021] In this embodiment, it should be specifically explained that a seat belt wear assessment model is constructed, and the seat belt usage records are imported into the seat belt wear assessment model to assess the wear of the seat belt during use. The safety assessment of the seat belt includes the following specific steps: S31. Obtain seat belt usage records, and substitute seat belt usage parameters into a seat belt wear index calculation formula to evaluate seat belt wear. The seat belt wear index calculation formula is: ,in, is the length of time the seat belt has been used, The maximum usage time of the seat belt, is the number of times the seat belt has been used, is the maximum number of times the seat belt is used, is the average task intensity level, that is, the average value of each task intensity level, is the maximum value of the task intensity level, is the time since the last maintenance, For maintenance cycle, 、 、 and is the weight, ; S32. Predict the safety belt wear index during use based on a deep learning neural network, obtain the roughness and hardness of the building surface, and at the same time obtain the contact area between the safety belt and the building surface, the safety belt wear index at the starting time, the operation time, the weight of the person and the climbing complexity, and construct a deep learning neural network whose input is the roughness and hardness of the building surface, the contact area between the safety belt and the building surface, the safety belt wear index at the starting time, the operation time, the weight of the person and the climbing complexity, and the output is the safety belt wear index during use. The specific steps are: obtain the roughness and hardness of the building surface during the historical use of the safety belt, and at the same time obtain the contact area between the safety belt and the building surface The contact surface of the building surface, the safety belt wear index at the starting time, the operation time, the weight and climbing complexity of the person, and the safety belt wear index before and after use are obtained. The historical data is divided into an 85% weighted and biased training set and a 15% weighted and biased test set. The 85% weighted and biased training set is input into the deep learning neural network model for training to obtain an initial deep learning neural network model. The initial deep learning neural network model is tested using the 15% weighted and biased test set, and the initial deep learning neural network model output that meets the maximum preset safety belt wear index accuracy during use is used as the deep learning neural network model. S33. Compare the seat belt wear index during use obtained by the neural network prediction with the seat belt threshold. If it is within the threshold range, the seat belt can continue to be used. If it exceeds the threshold range, replace the seat belt immediately and recalculate the seat belt wear index.
[0022] In this embodiment, it should be specifically explained that building a personnel risk assessment model and importing the basic information of the operating personnel into the personnel risk assessment model to assess the risk situation of the personnel themselves include the following specific steps: S41. Substitute the basic information parameters of the operator into the personnel risk calculation formula to evaluate the risk situation of the operator, wherein the personnel risk calculation formula is: ,in, The seat belt influence coefficient, P' is the seat belt wear index during use predicted by the neural network, P" is the seat belt threshold, and L is the working time of the operator since the start of work. is the shift length of the operator, BMI is the body mass index, 22 is the optimal BMI index, 13 is the maximum allowable body mass index deviation, and BMI=22 is the optimal health value. The greater the deviation, the higher the risk. The healthy BMI range is roughly 18-25, and the maximum allowable deviation is set to 13. E is the operator's age, which indicates the time the worker has been engaged in high-altitude work. C is the operator's climbing experience level, and the value range is generally set to 1 to 5. The coefficient 0.1 is used to control the speed at which experience growth reduces risk. The coefficient 2 emphasizes that professional experience level is more representative of high-altitude work ability than length of service. In many safety production scenarios, length of service may not necessarily equal high-altitude work ability, so the experience level is weighted and multiplied by 2 to highlight its impact on risk. The exponential decay rate is used to control the magnitude and rhythm of risk reduction due to experience accumulation. Too high or too low weight will affect work safety, such as difficulty in moving or poor physical strength.
[0023] In this embodiment, it should be specifically explained that constructing an environmental risk assessment model and importing environmental parameters of an operation period into the environmental risk assessment model to assess the operational risk caused by the environment includes the following specific steps: Substitute the environmental parameters of the operation period into the environmental risk assessment formula to evaluate the environmental conditions during the operation. The environmental risk assessment formula is: ,in, is the i-th environmental risk factor, is the weight, and n is the number of environmental risk factors, where environmental risk factors include temperature, light intensity, wind speed, etc. The environmental risk factor calculation formula is: ,in, is the i-th environmental parameter of the operation period, is the suitable value of the i-th environmental parameter.
[0024] In this embodiment, it should be specifically explained that the construction of a comprehensive risk assessment model, in which climbing complexity, safety belt wear index, personnel risk, and environmental risk are introduced into the comprehensive risk assessment model to evaluate the comprehensive safety of the operation scenario, and to provide an early warning includes the following specific steps: Substitute climbing complexity, personnel risk, and environmental risk into the comprehensive safety score formula for the operation scenario and obtain the comprehensive safety score of the operation scenario through weighted addition. Compare the comprehensive safety score of the operation scenario with the safety score threshold, set the warning level, and carry out normal operations if the safety level is safe. If the safety level is not safe, stop the operation; It should be noted here that the various setting parameters in this embodiment are obtained in the following manner: a representative seat belt wear index is obtained, and at the same time, a comprehensive safety score of the operating scenario is obtained, and experts are hired to manually judge whether the safety level meets the requirements. At the same time, the calculation results and judgment results of each step in this embodiment are substituted into the fitting software by the obtained historical data, and the values of the various setting parameters that meet the highest judgment accuracy are output.
[0025] The advantages of this embodiment over the prior art are: The present application obtains images of personnel working at a construction site, pre-processes the images to identify the personnel working conditions, obtains working environment parameters at the same time, constructs a climbing complexity assessment model, imports the collected personnel working conditions into the climbing complexity assessment model to evaluate the climbing complexity of the working area, constructs a safety belt wear assessment model, imports the safety belt usage records into the safety belt wear assessment model to evaluate the wear of the safety belt during use, evaluates the safety of the safety belt, constructs a personnel risk assessment model, imports the basic information of the operating personnel into the personnel risk assessment model to evaluate the personnel's own risk situation, constructs an environmental risk assessment model, imports the environmental parameters of the working period into the environmental risk assessment model to evaluate the working risk situation caused by the environment, constructs a comprehensive risk assessment model, imports the climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the working scene and issue early warnings, thereby improving the level of intelligent safety management at the construction site.
[0026] Example 2
[0027] like Figure 4 As shown, a deep learning-based construction site safety monitoring system is implemented based on the above-mentioned deep learning-based construction site safety monitoring method, which specifically includes a data acquisition module, a climbing complexity assessment module, a safety belt wear assessment module, a personnel risk assessment module, an environmental risk assessment module and a comprehensive risk assessment module. The data acquisition module is used to obtain images of personnel working on the construction site, pre-process the images to identify the personnel working conditions, and obtain the working environment parameters at the same time; the climbing complexity assessment module is used to assess the climbing complexity of the working area through the collected personnel working conditions; the safety belt wear assessment module is used to assess the wear of the safety belt during use through the use records of the safety belt, and assess the safety of the safety belt; the personnel risk assessment module is used to assess the risk of the personnel themselves through the basic information of the personnel; the environmental risk assessment module is used to assess the working risk caused by the environment through the environmental parameters of the working period; the comprehensive risk assessment module is used to assess the comprehensive safety situation of the working scene through climbing complexity, safety belt wear index, personnel risk and environmental risk, and issue an early warning.
[0028] Example 3
[0029] This embodiment provides an electronic device, comprising: a processor and a memory, wherein the memory stores a computer program that can be called by the processor; The processor executes the above-mentioned construction site safety monitoring method based on deep learning by calling the computer program stored in the memory.
[0030] The electronic device may vary significantly due to different configurations or performance, and may include one or more processors (Central Processing Units, CPUs) and one or more memories, wherein the memories store at least one computer program, which is loaded and executed by the processor to implement the deep learning-based construction site safety monitoring method provided in the above-mentioned method embodiment. The electronic device may also include other components for implementing the device's functions. For example, the electronic device may also have components such as wired or wireless network interfaces and input / output interfaces for data input and output. This embodiment will not be described in detail here.
[0031] Example 4
[0032] This embodiment provides a computer-readable storage medium having a rewritable computer program stored thereon; When the computer program runs on a computer device, the computer device executes the above-mentioned construction site safety monitoring method based on deep learning.
[0033] For example, computer-readable storage media can be read-only memory (ROM), random access memory (RAM), compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device.
[0034] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions according to the embodiments of the present application are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via a wired network or / and a wireless network. A computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.
Claims
1. A construction site safety monitoring method based on deep learning, characterized in that: It includes the following specific steps: Acquire images of personnel working at construction sites, pre-process the images to identify personnel working conditions, and simultaneously obtain working environment parameters; Construct a climbing complexity assessment model and import the collected personnel work conditions into the climbing complexity assessment model to evaluate the climbing complexity of the work area; Construct a seat belt wear assessment model, import seat belt usage records into the seat belt wear assessment model to evaluate the wear of seat belts during use and assess the safety of seat belts; Build a personnel risk assessment model and import the basic information of the operators into the personnel risk assessment model to assess the risk situation of the operators themselves; Construct an environmental risk assessment model and import the environmental parameters of the operation period into the environmental risk assessment model to evaluate the operational risks caused by the environment; Construct a comprehensive risk assessment model, import climbing complexity, safety belt wear index, personnel risk and environmental risk into the comprehensive risk assessment model to evaluate the comprehensive safety situation of the operation scenario and issue early warning.
2. The construction site safety monitoring method based on deep learning according to claim 1, characterized in that: The method of obtaining images of personnel working at a construction site, pre-processing the images to identify personnel working conditions, and obtaining working environment parameters comprises the following specific steps: S11. Using multiple high-definition cameras deployed in specific areas of the construction site, real-time images of workers climbing and moving on the structure are collected, and the collected images are pre-processed by frame separation, denoising, and enhancement. S12, identifying the location frame of the operator's area, cropping the entire frame image into a target area image, and using a mainstream human pose estimation algorithm to perform skeleton point recognition on the operator in the image frame to obtain a skeleton point set; S13. Obtain basic information of workers and environmental parameters of working hours and time periods.
3. The construction site safety monitoring method based on deep learning according to claim 2, characterized in that: The construction of the climbing complexity evaluation model and the importation of the collected personnel work conditions into the climbing complexity evaluation model to evaluate the climbing complexity of the work area include the following specific steps: S21. Obtain a set of trajectories of the operator's body center point in consecutive frames, and calculate the degree of trajectory tortuosity using a curvature variation coefficient, where the curvature variation coefficient is the ratio of the total length of the trajectory broken line segments to the straight-line distance between the starting point and the end point. A height fluctuation index is obtained by calculating the standard deviation of the height of the body center point per unit time. S22. Calculate the climbing speed of each frame based on the displacement vector of the skeleton center point in different frames to obtain a climbing speed sequence. Substitute the climbing speed into the speed fluctuation standard deviation calculation formula to evaluate the operator's climbing speed fluctuation by calculating the speed standard deviation. S23. Segment the background and obstacles of each image frame using a semantic segmentation model, identify obstructing and interfering structures on the climbing path, and evaluate the interference of obstacles using an obstacle occlusion ratio, where the obstacle occlusion ratio is the ratio of the intersection of the obstacle area in the image and the climbing path area predicted by the skeleton to the climbing path area predicted by the skeleton; S24. Substitute the curvature variation coefficient, height fluctuation index, speed fluctuation standard deviation, and obstacle shielding ratio into the climbing complexity calculation formula to evaluate the complexity of the climbing behavior of the person in a specific time period. The climbing complexity calculation formula is: , where the curvature variation coefficient is F, the height fluctuation index is H, the speed fluctuation standard deviation is V, and the obstacle occlusion ratio is D. 、 、 and is the weight.
4. The construction site safety monitoring method based on deep learning according to claim 3 is characterized in that: The construction of the seat belt wear assessment model, importing the seat belt usage records into the seat belt wear assessment model to assess the wear of the seat belt during use, and assessing the safety of the seat belt includes the following specific steps: S31. Obtain seat belt usage records, and substitute seat belt usage parameters into a seat belt wear index calculation formula to evaluate seat belt wear. The seat belt wear index calculation formula is: ,in, is the length of time the seat belt has been used, The maximum usage time of the seat belt, is the number of times the seat belt has been used, is the maximum number of times the seat belt is used, is the average task intensity level, is the maximum value of the task intensity level, is the time since the last maintenance, For maintenance cycle, 、 、 and is the weight, ; S32. Predicting the seat belt wear index during use based on a deep learning neural network, obtaining the roughness and hardness of the building surface, and simultaneously obtaining the contact area between the seat belt and the building surface, the seat belt wear index at the start time, the operation duration, the weight of the person, and the climbing complexity. Constructing a deep learning neural network whose inputs are the roughness and hardness of the building surface, the contact area between the seat belt and the building surface, the seat belt wear index at the start time, the operation duration, the weight of the person, and the climbing complexity, and whose output is the seat belt wear index during use; S33. Compare the seat belt wear index during use obtained by the neural network prediction with the seat belt threshold. If it is within the threshold range, the seat belt can continue to be used. If it exceeds the threshold range, replace the seat belt immediately and recalculate the seat belt wear index.
5. The construction site safety monitoring method based on deep learning according to claim 4, characterized in that: The construction of the personnel risk assessment model and the importation of the basic information of the operating personnel into the personnel risk assessment model to assess the risk situation of the personnel themselves include the following specific steps: S41. Substitute the basic information parameters of the operator into the personnel risk calculation formula to evaluate the risk situation of the operator, wherein the personnel risk calculation formula is: ,in, The seat belt influence coefficient, P' is the seat belt wear index during use predicted by the neural network, P" is the seat belt threshold, and L is the working time of the operator since the start of work. is the operator's shift length, BMI is the body mass index, E is the operator's age, and C is the operator's climbing experience level.
6. The construction site safety monitoring method based on deep learning according to claim 5, characterized in that: The construction of the environmental risk assessment model and the importation of the environmental parameters of the operation period into the environmental risk assessment model to assess the operation risk caused by the environment include the following specific steps: Substitute the environmental parameters of the operation period into the environmental risk assessment formula to evaluate the environmental conditions during the operation. The environmental risk assessment formula is: ,in, is the i-th environmental risk factor, is the weight, n is the number of environmental risk factors, where environmental risk factors include temperature, light intensity and wind speed. The environmental risk factor calculation formula is: ,in, is the i-th environmental parameter of the operation period, is the suitable value of the i-th environmental parameter.
7. The construction site safety monitoring method based on deep learning according to claim 6, characterized in that: The construction of a comprehensive risk assessment model, which incorporates climbing complexity, safety belt wear index, personnel risk, and environmental risk into the comprehensive risk assessment model to assess the overall safety of the operation scenario and provide early warning, includes the following specific steps: Substitute climbing complexity, personnel risk, and environmental risk into the comprehensive safety score formula for the operation scenario and obtain the comprehensive safety score for the operation scenario through weighted addition. Compare the comprehensive safety score for the operation scenario with the safety score threshold, set the warning level, and perform normal operations for the safety level. Stop operations for non-safety levels.
8. A construction site safety monitoring system based on deep learning, which is implemented based on the construction site safety monitoring method based on deep learning according to any one of claims 1 to 7, characterized in that: Specifically include: The data acquisition module is used to obtain images of personnel working at the construction site, pre-process the images to identify the personnel working conditions, and obtain working environment parameters; The climbing complexity assessment module is used to assess the climbing complexity of the work area based on the collected personnel work conditions; The seat belt wear assessment module is used to evaluate the wear of seat belts during use and the safety of seat belts through seat belt usage records; The personnel risk assessment module is used to assess the risk situation of personnel themselves based on their basic information; Environmental risk assessment module, used to evaluate the operational risks caused by the environment through environmental parameters during the operation period; The comprehensive risk assessment module is used to evaluate the overall safety situation of the operation scenario through climbing complexity, safety belt wear index, personnel risk and environmental risk, and provide early warning.
9. An electronic device comprising: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; It is characterized in that the processor executes the construction site safety monitoring method based on deep learning as described in any one of claims 1 to 7 by calling the computer program stored in the memory.
10. A computer-readable storage medium, characterized in that Instructions are stored, and when the instructions are run on a computer, the computer executes the construction site safety monitoring method based on deep learning as described in any one of claims 1 to 7.