Work clothes detection method, system, equipment and medium combining yolov8 and target tracking

Abstract

The present application discloses a work clothes detection method, system, equipment and medium combining yolov8 and target tracking, which relates to the field of safety detection technology. The method comprises acquiring surveillance video data of a target monitoring area and performing data preprocessing; performing work clothes target detection on the preprocessed surveillance video data in a preset work clothes detection model to obtain a work clothes detection result; performing target tracking analysis based on the work clothes detection result to obtain a work clothes tracking result of a continuous video stream; constructing a preset work clothes detection model in combination with the yolov8 algorithm and the target tracking algorithm; generating a defect detection mark based on the work clothes detection result and triggering an alarm instruction; acquiring geographic location information based on the work clothes tracking result, performing spatial matching analysis based on the geographic location information and the work clothes tracking result to obtain a safety target detection result of the corresponding target monitoring area; the present application provides a deep analysis method combining computer vision and deep learning technology to improve the efficiency of safety monitoring.

Description

Technical Field

[0001] The present application relates to the field of safety detection technology, and in particular to a work clothes detection method, system, equipment and medium combining yolov8 and target tracking. Background Art

[0002] In the power industry, the importance of safety monitoring is self-evident, as it is directly related to the stable operation of power facilities and the safety of personnel. However, traditional manual monitoring methods are not only inefficient but also prone to interference from human factors, making it difficult to ensure the continuity and accuracy of monitoring.

[0003] With the development of artificial intelligence technology, how to use computer vision, deep learning and other technologies to deeply analyze video data and provide a new security monitoring solution for the power industry has always been a technical research direction that has attracted much attention in the field of power safety. Summary of the Invention

[0004] In order to provide a deep analysis method that combines computer vision and deep learning technology to improve the efficiency of safety monitoring, this application provides a work clothes detection method, system, equipment and medium that combines yolov8 and target tracking.

[0005] In the first aspect, the invention objectives of this application are achieved by adopting the following technical solutions:

[0006] A work clothes detection method combining yolov8 and target tracking, including:

[0007] Obtain surveillance video data of the target surveillance area and perform data preprocessing;

[0008] In a preset work clothes detection model, work clothes target detection is performed on the pre-processed surveillance video data to obtain a work clothes detection result; based on the work clothes detection result, target tracking analysis is performed to obtain a work clothes tracking result of a continuous video stream; the preset work clothes detection model is constructed by combining the yolov8 algorithm and the target tracking algorithm;

[0009] generating a defect detection mark based on the work clothes inspection result and triggering an alarm instruction;

[0010] The geographic location information is obtained based on the work uniform tracking result, and a spatial matching analysis is performed based on the geographic location information and the work uniform tracking result to obtain a safety target detection result corresponding to the target monitoring area.

[0011] By adopting the above technical solution, surveillance video data is obtained based on computer vision, and the preprocessing of surveillance video data includes denoising, normalization and frame extraction processing, which improves the quality of video data. The work clothes target detection of this application includes people wearing safety helmets and people wearing specific electric work clothes. It detects whether the personnel are wearing safety helmets and whether the electric workers are wearing specific electric work clothes correctly. The work clothes detection model can distinguish between electric work clothes and other types of work clothes or ordinary clothing. This classification detection capability is crucial to ensure that workers comply with the dress code regulations of the power industry and helps to improve the overall safety level of the work site. The yolov8 algorithm is used for work clothes detection. The yolov8 algorithm has high generalization ability and flexibility. It can identify whether workers are wearing safety helmets and specific work clothes correctly in complex environments such as production workshops in the power industry. At the same time, the high accuracy of the yolov8 algorithm reduces the possibility of false alarms and missed alarms, and improves Reliability and efficiency of detection; combined with the target tracking algorithm, the detected target is continuously tracked, its movement trajectory and state changes are recorded, effectively avoiding multiple repeated alarms, improving the accuracy and effectiveness of the alarm, and responding to potential safety risks in a timely manner. Through multi-threading technology and frame skipping technology, the processing speed of target tracking is improved and resource usage is reduced, ensuring the efficient operation of the system; then, based on the work clothes defect detection results, defect detection marks are generated, and alarm instructions are triggered, so that the system can take immediate action when safety violations are detected, improving the timeliness and effectiveness of safety monitoring; based on the work clothes defect tracking results, geographic location information is obtained, and spatial matching analysis is performed to obtain safety defect target detection results for the corresponding target monitoring area, so that the system can more accurately locate the location of the safety violation, so that the present application provides a deep analysis method that combines computer vision and deep learning technology to improve safety monitoring efficiency.

[0012] In a preferred embodiment of the present application, the acquisition of surveillance video data of the target surveillance area and data preprocessing specifically include:

[0013] Performing frame processing on the surveillance video data of the target surveillance area to obtain multiple frames of continuous surveillance images to be determined;

[0014] The method includes: performing work clothes target detection on preprocessed surveillance video data in a preset work clothes detection model to obtain a work clothes detection result; performing target tracking analysis based on the work clothes detection result to obtain a work clothes tracking result of a continuous video stream, including: performing work clothes target detection on multiple frames of the surveillance images to be determined frame by frame based on the yolov8 algorithm in the preset work clothes detection model to obtain initial work clothes detection results of the corresponding continuous multiple frames of surveillance images; the initial work clothes detection results include a number of work clothes detection targets;

[0015] In a preset work clothes detection model, target tracking analysis is performed on several work clothes detection targets based on the target tracking algorithm and the initial attack detection results to obtain a work clothes tracking result of a continuous video stream.

[0016] By adopting the above technical solution, the monitoring video data is frame-processed to ensure the continuity and integrity of the monitoring video data. In the preset work clothes detection model, the work clothes target detection is performed frame by frame on multiple frames of monitoring images to be judged based on the yolov8 algorithm to obtain the initial work clothes detection results of the corresponding continuous multiple frames of monitoring images. The yolov8 algorithm has high generalization ability and flexibility, and can accurately identify whether the staff are correctly wearing safety helmets and specific work clothes in various complex environments; at the same time, the high accuracy of the yolov8 algorithm reduces the possibility of false alarms and missed alarms. The target tracking algorithm can continuously track the detected target, record its motion trajectory and state changes, avoid multiple repeated alarms, and improve the accuracy and effectiveness of the alarm.

[0017] In a preferred example of the present application, the geographic location information is obtained based on the work uniform tracking result, and spatial matching analysis is performed based on the geographic location information and the work uniform tracking result to obtain a safety target detection result corresponding to the target monitoring area, specifically including:

[0018] Obtaining the geographical location coordinates of the detection device of the monitoring video data, and performing plane coordinate projection conversion on the geographical location coordinates to obtain plane rectangular coordinates;

[0019] Based on the plane rectangular coordinates, when performing frame processing on the monitoring video data, the image coordinates of each frame of the monitoring image are positioned using a linear interpolation method to obtain frame image positioning information;

[0020] Determining the geographic location information of the monitoring image of the work clothes tracking result based on the plane rectangular coordinate system and the frame image positioning information;

[0021] Tracking monitoring is based on the geographic location information of the same work clothes detection target in multiple work clothes tracking results of continuous video streams to obtain the safety target detection results of the corresponding target monitoring area.

[0022] By adopting the above technical solution, the geographic location coordinates of the detection equipment (such as a camera) of the monitoring video data are obtained to obtain accurate geographic location information, which provides a basis for subsequent spatial matching analysis; the obtained geographic location coordinates are then transformed into plane coordinate projection to facilitate subsequent calculations and processing, so as to accurately correspond the geographic information with the location information in the monitoring image, thereby improving the accuracy of the geographic location information; when the monitoring video is framed, each frame of the image is used as an independent processing unit to ensure that each frame of the image can be processed independently, and it is convenient to perform frame detection analysis during subsequent target tracking analysis, which is beneficial to reduce the amount of model calculation; at the same time, the image coordinates of each frame of the monitoring image are positioned based on linear interpolation to obtain frame image positioning information, and the coordinates of the unknown points are calculated through the known coordinate points to ensure the accuracy of the image coordinates, thereby improving the accuracy of the image coordinate positioning, and based on the plane rectangular coordinate system and the frame image positioning information, the geographic location information of the monitoring image of the work clothes tracking result is determined, and the detection target position in the image is matched with the position in the plane rectangular coordinate system to ensure the accuracy of the spatial matching.

[0023] In a preferred embodiment of the present application, after acquiring the surveillance video data of the target surveillance area and performing data preprocessing, the method further includes:

[0024] Determine, based on the monitoring video data, a first predicted position of the work clothes detection target when it enters the monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area;

[0025] generating a detection target tracking comparison table for adjusting the accuracy of work clothes detection target tracking based on the monitoring video data, the first predicted position, and the second predicted position;

[0026] Obtaining a target position change reference range representing a target position change threshold for work clothes detection, and performing adaptive parameter optimization on the target tracking algorithm based on the target position change reference range and the detection target tracking comparison table;

[0027] The position information of a first actual work clothes detection target is obtained, and the position information of the first actual work clothes detection target is input into the preset work clothes detection model to adjust the tracking accuracy of the work clothes detection target.

[0028] By adopting the above technical solution, the first predicted position is to predict the position of the target when it enters the monitoring area by analyzing the motion trajectory and target features in the monitoring video data. The second predicted position is to judge when the work clothes detection target passes through a specific area through the monitoring video data, and to predict the position of the target when it passes through a specific area by analyzing the motion trajectory and residence time of the target in the specific area. By accurately predicting the position of the target, it is possible to prepare in advance and improve the efficiency and accuracy of detection and tracking; the detection target tracking comparison table contains the position information and corresponding tracking parameters of the target in different time periods, which provides a reference basis for subsequent tracking accuracy adjustment, ensuring that the system can dynamically adjust the tracking parameters according to actual conditions, and improve the tracking accuracy and stability. The position change threshold is used to determine whether the change in the target position exceeds the normal range. Through adaptive parameter optimization, the tracking parameters are dynamically adjusted according to the actual movement of the detection target to improve the robustness and adaptability of tracking. At the same time, the parameters of the tracking algorithm are dynamically adjusted according to the actual position information and the parameters in the comparison table to ensure that the tracking accuracy is always in the optimal state.

[0029] In a preferred example of the present application: the monitoring video data includes the camera type, the camera installation height, the distance between the camera and the monitoring area, and the width and height of the monitoring area;

[0030] Determining, based on the monitoring video data, a first predicted position of a work clothes detection target when it enters a monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area, specifically includes:

[0031] Obtain parameter information of the work clothes detection model, including the detection accuracy of the yolov8 algorithm, the tracking accuracy of the target tracking algorithm, and the minimum and maximum sizes of the work clothes detection target;

[0032] Determining a first predicted position of the work clothes detection target when it enters the monitoring area according to the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target;

[0033] The second predicted position of the work clothes detection target when passing through a specific area is determined according to the camera type, the width and height of the monitoring area, the maximum size of the work clothes detection target, and the tracking accuracy of the target tracking algorithm.

[0034] By adopting the above technical solution, detailed environmental parameters are obtained based on the physical settings or configuration parameters of the camera, providing accurate environmental parameters for subsequent position prediction. Detailed work clothes detection model parameter information provides detailed information on the algorithm and target features for the predicted position, thereby improving the accuracy of position prediction. Through precise geometric calculations and target feature analysis, the system can accurately predict the position of the work clothes detection target when it enters the monitoring area, thereby improving the timeliness and accuracy of detection. By comprehensively considering the camera type, monitoring area size, target features and tracking accuracy, the system can accurately predict the position of the work clothes detection target when it passes through a specific area, thereby improving the accuracy and reliability of tracking.

[0035] In a preferred example of the present application, generating a detection target tracking comparison table for adjusting the tracking accuracy of the work clothes detection target based on the monitoring video data, the first predicted position, and the second predicted position specifically includes: obtaining a predicted trajectory length of the work clothes detection target based on the width and height of the monitoring area, the first predicted position, and the second predicted position;

[0036] According to the detection accuracy of the yolov8 algorithm and the tracking accuracy of the target tracking algorithm, the predicted detection time and the predicted tracking time of the work clothes detection target are obtained;

[0037] Obtaining a predicted tracking speed of the work clothes detection target according to the predicted trajectory length of the work clothes detection target, the predicted detection time, and the predicted tracking time;

[0038] Calculating a displacement distance between the first predicted position and the second predicted position to obtain a predicted displacement ratio for predicting an actual displacement distance;

[0039] Calculate the detection and tracking time difference between the predicted detection time and the predicted tracking time, and calculate the tracking speed difference between the predicted tracking speed and the actual tracking speed of the work clothes detection target;

[0040] Obtaining a predicted tracking time correspondence for determining an actual tracking speed of a work clothes detection target based on the detection tracking time difference and the tracking speed difference;

[0041] According to the predicted displacement ratio and the predicted tracking speed-time correspondence, a detection target tracking comparison table for comparing and adjusting the tracking accuracy of the work clothes detection target is generated.

[0042] By adopting the above technical solution, a work clothes detection method combining computer vision and deep learning technology is provided, which can accurately predict and track work clothes detection targets, improve the efficiency and reliability of safety monitoring, and can geometrically calculate and accurately obtain the predicted trajectory length of the work clothes detection target to provide basic data for subsequent tracking accuracy adjustment. Through experimental data and model parameters, the detection time of the yolov8 algorithm and the tracking time of the target tracking algorithm can be accurately calculated to improve the prediction accuracy and stability, and through the time-distance relationship, the predicted tracking speed of the target can be accurately calculated; through geometric calculation, the system can accurately calculate the displacement ratio of the target; through the calculation of time difference and speed difference, the system can better adjust the tracking parameters and improve the tracking accuracy and stability. Through the relationship between time difference and speed difference, the system can establish a corresponding relationship between predicted tracking time and actual tracking time, thereby improving the tracking accuracy and real-time performance.

[0043] In a preferred embodiment of the present application, before determining the first predicted position of the work clothes detection target when it enters the monitoring area based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, the method further includes:

[0044] Obtaining a camera adjustment angle, and determining recognition accuracy of a work clothes detection target based on the camera installation height and the camera adjustment angle;

[0045] Obtaining a first coefficient for adjusting the first predicted position according to the recognition accuracy of the work clothes detection target;

[0046] The determining, based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, of a first predicted position when the work clothes detection target enters the monitoring area specifically includes:

[0047] A first predicted position of the work clothes detection target entering the monitoring area is determined based on the first coefficient, the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target.

[0048] By adopting the above technical solution, the recognition accuracy of the work clothes detection target is judged according to the camera installation height and the camera adjustment angle. Through geometric calculation and experimental data, the influence of the camera adjustment angle on the recognition accuracy is analyzed to more accurately evaluate the recognition accuracy of the work clothes detection target and improve the accuracy of detection; by obtaining the first coefficient, the system can dynamically adjust the first predicted position according to the recognition accuracy; by comprehensively considering the camera adjustment angle, recognition accuracy and target characteristics, the system can more accurately predict the position of the work clothes detection target when it enters the monitoring area.

[0049] In the second aspect, the invention objective of this application is achieved by adopting the following technical solutions:

[0050] A work clothes detection system combining YOLOv8 and target tracking is applied to the above-mentioned work clothes detection method combining YOLOv8 and target tracking, and the system includes:

[0051] Data acquisition module, used to obtain surveillance video data of the target surveillance area and perform data preprocessing;

[0052] A work clothes detection module is used to perform work clothes target detection on the pre-processed monitoring video data in a preset work clothes detection model to obtain a work clothes detection result; the preset work clothes detection model is constructed by combining the YOLOv8 algorithm and the target tracking algorithm;

[0053] A target tracking module, configured to perform target tracking analysis based on the work clothes detection results to obtain work clothes tracking results of a continuous video stream;

[0054] A defect detection module, configured to generate a defect detection mark based on the work clothes detection result and trigger an alarm instruction;

[0055] The geographic location module is used to obtain geographic location information based on the work uniform tracking result, perform spatial matching analysis based on the geographic location information and the work uniform tracking result, and obtain a safety target detection result corresponding to the target monitoring area.

[0056] In a third aspect, the invention objective of this application is achieved by adopting the following technical solutions:

[0057] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned work clothes detection method combining YOLOv8 and target tracking are implemented.

[0058] Fourthly, the invention objectives of this application are achieved by adopting the following technical solutions:

[0059] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned work clothes detection method combining yolov8 and target tracking.

[0060] In summary, this application includes at least one of the following beneficial technical effects:

[0061] 1. Use the yolov8 algorithm to detect work clothes. The yolov8 algorithm has high generalization ability and flexibility, and can identify whether workers are wearing safety helmets and specific work clothes correctly in complex environments such as production workshops in the power industry. At the same time, the high accuracy of the yolov8 algorithm reduces the possibility of false alarms and missed alarms, and improves the reliability and efficiency of detection; combined with the target tracking algorithm, the detected target is continuously tracked, and its motion trajectory and state changes are recorded, effectively avoiding multiple repeated alarms, improving the accuracy and effectiveness of the alarm, and responding to potential safety risks in a timely manner. Through multi-threading technology and frame skipping technology, the processing speed of target tracking is improved and resource usage is reduced, ensuring the efficient operation of the system; then, based on the work clothes defect detection results, a defect detection mark is generated, and an alarm instruction is triggered, so that the system can take immediate action when a safety violation is detected, thereby improving the timeliness and effectiveness of safety monitoring; based on the work clothes defect tracking results, geographic location information is obtained, and spatial matching analysis is performed to obtain the safety defect target detection results of the corresponding target monitoring area, so that the system can more accurately locate the location of the safety violation, so that the present application provides a deep analysis method that combines computer vision and deep learning technology to improve the efficiency of safety monitoring;

[0062] 2. A work clothes detection method combining computer vision and deep learning technology is provided, which can accurately predict and track work clothes detection targets, improve the efficiency and reliability of safety monitoring, and can geometrically calculate the predicted trajectory length of the work clothes detection target by accurately obtaining it, providing basic data for subsequent tracking accuracy adjustment. Through experimental data and model parameters, the detection time of the yolov8 algorithm and the tracking time of the target tracking algorithm can be accurately calculated to improve the prediction accuracy and stability, and the predicted tracking speed of the target can be accurately calculated through the time-distance relationship; through geometric calculation, the system can accurately calculate the displacement ratio of the target; through the calculation of time difference and speed difference, the system can better adjust the tracking parameters and improve the tracking accuracy and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of a work clothes detection method combining yolov8 and target tracking in one embodiment of the present application;

[0064] Figure 2 This is a flowchart of step S4 in a work clothes detection method combining yolov8 and target tracking in one embodiment of the present application;

[0065] Figure 3 It is a schematic diagram of a device in one embodiment of the present application. DETAILED DESCRIPTION

[0066] The present application is further described in detail below with reference to the accompanying drawings.

[0067] In one embodiment, if Figure 1 As shown, this application discloses a work clothes detection method combining yolov8 and target tracking, which specifically includes the following steps:

[0068] S1: Obtain surveillance video data of the target surveillance area and perform data preprocessing.

[0069] In this embodiment, the target monitoring area refers to a specific area where work clothes inspection is required, such as a power construction site; data preprocessing includes operations such as denoising, normalization, and resolution adjustment on the monitoring video data.

[0070] S2: In the preset work clothes detection model, work clothes target detection is performed on the preprocessed surveillance video data to obtain the work clothes detection results; target tracking analysis is performed based on the work clothes detection results to obtain the work clothes tracking results of the continuous video stream; the preset work clothes detection model is constructed in combination with the yolov8 algorithm and the target tracking algorithm.

[0071] In this embodiment, the work clothes detection model is a model that combines the yolov8 algorithm and the target tracking algorithm, and is used to detect and track work clothes targets; work clothes target detection refers to the use of the yolov8 algorithm to perform work clothes target detection on the pre-processed surveillance video data, and identify people wearing safety helmets and specific electrical work clothes; work clothes tracking results refer to the monitoring video data based on the specified frame extraction frequency, after frame processing, used to continuously track the detected target, and record the target's application trajectory and state changes. Tracking data set.

[0072] Specifically, the yolov8 algorithm is used to perform work clothes target detection on the preprocessed surveillance video data, identify people wearing safety helmets and specific electrical work clothes, and obtain work clothes detection results; the target tracking algorithm is used to continuously track the detected target, record the target's motion trajectory and state changes, and obtain work clothes tracking results.

[0073] S3: Generate defect detection marks based on the work clothes inspection results and trigger alarm instructions.

[0074] In this embodiment, the defect detection mark refers to the mark information generated when a person is detected not wearing work clothes as required, and the alarm instruction refers to the warning notification sent by the system when a violation of work clothes regulations is detected.

[0075] Specifically, when a person is detected not wearing the required uniform, a defect detection mark is generated, including the time, location, and type of uniform violation. The marked image is saved as evidence for subsequent on-site monitoring, record archiving, and analysis. Alarm notifications are sent to management personnel via sound, light, or network messages, and the alarm information and marked image are uploaded to a server, enabling information sharing and remote monitoring.

[0076] S4: Obtain geographic location information based on the work uniform tracking results, perform spatial matching analysis based on the geographic location information and the work uniform tracking results, and obtain the safety target detection results of the corresponding target monitoring area.

[0077] In this embodiment, geographic location information refers to the geographic location coordinates of the detection device (such as a camera) of the monitoring video data corresponding to the work clothes tracking results; spatial matching analysis refers to matching and analyzing the geographic location information with the work clothes tracking results to determine the actual location of the target.

[0078] Specifically, when the monitoring video data is framed, the linear interpolation method is used to locate the image coordinates of each frame of the monitoring image to obtain the frame image positioning information. Based on the plane rectangular coordinate system and the frame image positioning information, the geographical location information of the monitoring image of the work clothes tracking result is determined. The tracking and monitoring is based on the geographical location information of multiple work clothes tracking results of the same work clothes detection target in the continuous video stream to obtain the safety target detection result of the corresponding target monitoring area.

[0079] In one embodiment, in step S1, surveillance video data of a target surveillance area is acquired and data preprocessing is performed, specifically including:

[0080] S11: performing frame processing on the surveillance video data of the target surveillance area to obtain multiple frames of continuous surveillance images to be determined.

[0081] In this embodiment, a video processing library (such as OpenCV) is used to segment video data into multiple frames of images.

[0082] In this embodiment, in step S2, work clothes target detection is performed on the preprocessed surveillance video data in a preset work clothes detection model to obtain a work clothes detection result; target tracking analysis is performed based on the work clothes detection result to obtain a work clothes tracking result of the continuous video stream, including:

[0083] S21: In the preset work clothes detection model, work clothes target detection is performed frame by frame on multiple frames of surveillance images to be judged based on the yolov8 algorithm to obtain initial work clothes detection results of the corresponding continuous multiple frames of surveillance images; the initial work clothes detection results include several work clothes detection targets.

[0084] Specifically, use the work clothes detection model, load the pre-trained weight file, perform target detection on each frame image, and output the detection box and category label.

[0085] S22: In a preset work clothes detection model, target tracking analysis is performed on several work clothes detection targets based on the target tracking algorithm and the initial attack detection results to obtain a work clothes tracking result of a continuous video stream.

[0086] In this embodiment, a target tracking algorithm (such as a Kalman filter or an optical flow method) is used to track and analyze several work clothes detection targets in the initial work clothes detection results, and the motion trajectory of each detection target is recorded, including position, speed and direction.

[0087] In one embodiment, if Figure 2 As shown, in step S4, the geographic location information is obtained based on the work uniform tracking result, and a spatial matching analysis is performed based on the geographic location information and the work uniform tracking result to obtain the safety target detection result of the corresponding target monitoring area, specifically including:

[0088] S41: Obtain the geographical location coordinates of the detection device of the monitoring video data, and perform plane coordinate projection conversion on the geographical location coordinates to obtain plane rectangular coordinates.

[0089] Specifically, plane coordinate projection conversion refers to converting geographic coordinates (latitude and longitude) into coordinates in a plane rectangular coordinate system; the latitude and longitude coordinates of the camera are obtained through the camera's built-in GPS module or other external positioning devices.

[0090] S42: Based on plane rectangular coordinates, when performing frame processing on the monitoring video data, linear interpolation is used to locate the image coordinates of each frame of the monitoring image to obtain frame image positioning information.

[0091] Specifically, the framed image positioning information refers to the location information of the target in each frame of the image, and a map projection algorithm (such as UTM projection or Mercator projection) is used to convert the latitude and longitude coordinates into plane rectangular coordinates.

[0092] S43: Determine the geographical location information of the monitoring image of the work clothes tracking result based on the plane rectangular coordinate system and the frame image positioning information.

[0093] In this embodiment, the target position information in each frame of image is matched with the position information in the plane rectangular coordinate system to determine the actual geographical location of the target, and the geographical location information of each target at different time points is recorded.

[0094] S44: Tracking monitoring obtains a safety target detection result corresponding to the target monitoring area based on the geographic location information of the same work clothes detection target in multiple work clothes tracking results of the continuous video stream.

[0095] In this embodiment, tracking monitoring is based on multiple work clothes tracking results of the same work clothes detection target in a continuous video stream, recording the target's movement trajectory, judging whether the target is in a safe area based on the target's geographic location information and movement trajectory, and generating a safe target detection result. If it is detected that the target has left the safe area or there are other safety hazards, an alarm notification is triggered.

[0096] In one embodiment, after step S1, the work clothes detection method combining yolov8 and target tracking further includes:

[0097] S101: Determine, based on monitoring video data, a first predicted position of a work clothes detection target when it enters a monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area.

[0098] In this embodiment, the monitoring video data includes the camera type, the camera installation height, the distance between the camera and the monitoring area, and the width and height of the monitoring area.

[0099] Specifically, the detection accuracy of the yolov8 algorithm, the tracking accuracy of the target tracking algorithm, and the minimum and maximum sizes of the work clothes detection target are obtained from the model training and experimental data. Using geometric calculation and target feature analysis, the position of the target when it enters the monitoring area is calculated according to the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, and the first predicted position is obtained; the position of the target when it passes through a specific area is calculated using the camera type, the width and height of the monitoring area, the maximum size of the work clothes detection target, and the tracking accuracy of the target tracking algorithm, and the second predicted position is obtained.

[0100] In this embodiment, determining the first predicted position of the work clothes detection target when it enters the monitoring area and the second predicted position of the work clothes detection target when it passes through a specific area based on the monitoring video data specifically includes:

[0101] S1011: Obtain parameter information of the work clothes detection model, including the detection accuracy of the yolov8 algorithm, the tracking accuracy of the target tracking algorithm, and the minimum size and maximum size of the work clothes detection target.

[0102] S1012: Determine a first predicted position of the work clothes detection target when it enters the monitoring area based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target.

[0103] In this embodiment, the distance between the camera and the monitoring area refers to the straight-line distance between the camera and the center point of the monitoring area, and the minimum size of the work clothes detection target refers to the minimum visible size of the work clothes detection target in the image.

[0104] Example: The horizontal position of the target when it enters the monitoring area (the first predicted position at this time) is equal to the distance between the camera and the monitoring area multiplied by the minimum size of the work clothes detection target, and then divided by the camera installation height.

[0105] S1013: Determine a second predicted position of the work clothes detection target when it passes through the specific area based on the camera type, the width and height of the monitoring area, the maximum size of the work clothes detection target, and the tracking accuracy of the target tracking algorithm.

[0106] In this embodiment, the width and height of the monitoring area refer to the horizontal and vertical dimensions of the monitoring area; the maximum size of the work clothes detection target refers to the maximum visible size of the work clothes detection target in the image; and the tracking accuracy of the target tracking algorithm refers to the accuracy of the target tracking algorithm when tracking the work clothes target.

[0107] Example: The horizontal position of the target when it passes through a specific area (the second predicted position at this time) is equal to the width of the monitoring area multiplied by the maximum size of the work clothes detection target, divided by the tracking accuracy of the target tracking algorithm.

[0108] S102: Generate a detection target tracking comparison table for comparing and adjusting the work clothes detection target tracking accuracy based on the monitoring video data, the first predicted position, and the second predicted position.

[0109] In this embodiment, a detection target tracking comparison table for adjusting the accuracy of work clothes detection target tracking is generated based on the monitoring video data, the first predicted position, and the second predicted position, specifically including:

[0110] S1021: Obtain the predicted trajectory length of the work clothes detection target according to the width and height of the monitoring area, the first predicted position and the second predicted position.

[0111] Example: Predicted trajectory length for work clothes detection targets (x1, y1) and (x2, y2) are the coordinates of the first predicted position and the second predicted position respectively.

[0112] S1022: Obtain the predicted detection time and predicted tracking time of the work clothes detection target based on the detection accuracy of the yolov8 algorithm and the tracking accuracy of the target tracking algorithm.

[0113] In this embodiment, the predicted detection time and the predicted tracking time refer to the detection time of the yolov8 algorithm and the tracking time of the target tracking algorithm, respectively.

[0114] S1023: Obtain a predicted tracking speed of the work clothes detection target according to the predicted trajectory length, predicted detection time, and predicted tracking time of the work clothes detection target.

[0115] In this embodiment, the predicted tracking speed refers to the expected moving speed of the work clothes detection target.

[0116] S1024: Calculate the displacement distance between the first predicted position and the second predicted position to obtain a predicted displacement ratio for predicting the actual displacement distance.

[0117] In this embodiment, the predicted displacement ratio refers to the ratio of the displacement distance between the first predicted position and the second predicted position.

[0118] S1025: Calculate the detection and tracking time difference between the predicted detection time and the predicted tracking time, and calculate the tracking speed difference between the predicted tracking speed and the actual tracking speed of the work clothes detection target.

[0119] In this embodiment, the detection tracking time difference and the tracking speed difference refer to the difference between the predicted detection time and the actual detection time, and the difference between the predicted tracking speed and the actual tracking speed, respectively.

[0120] S1026: Obtaining a predicted tracking time correspondence for determining an actual tracking speed of the work clothes detection target according to the detection tracking time difference and the tracking speed difference.

[0121] In this embodiment, based on the detection tracking time difference and the tracking speed difference, the predicted tracking time correspondence for judging the actual tracking speed of the work clothes detection target is obtained. In actual application, a linear regression or neural network model can be used to establish the relationship between the time difference and the speed difference based on historical data.

[0122] S1027: Generate a detection target tracking comparison table for comparing and adjusting the tracking accuracy of the work clothes detection target according to the corresponding relationship between the predicted displacement ratio and the predicted tracking speed time.

[0123] Specifically, a comparison table is generated based on the corresponding relationship between the predicted displacement ratio and the predicted tracking speed time, and the predicted and actual data under different situations are recorded to facilitate the subsequent adjustment of the parameters of the tracking algorithm.

[0124] S103: Obtain a target position change reference range representing a target position change threshold for work clothes detection, and perform adaptive parameter optimization on a target tracking algorithm according to the target position change reference range and a detection target tracking comparison table.

[0125] Specifically, the target position change reference range refers to the reasonable range of target position change determined through experimental data and historical records; according to the target position change reference range and the data in the comparison table, the parameters of the tracking algorithm, such as the tracking window size, search range, update frequency, etc., are dynamically adjusted.

[0126] S104: Acquire position information of a first actual work clothes detection target, and input the position information of the first actual work clothes detection target into a preset work clothes detection model to adjust the tracking accuracy of the work clothes detection target.

[0127] Specifically, through real-time monitoring and target tracking algorithms, the precise location information of the current work clothes detection target is obtained; the location information of the first actual work clothes detection target is input into the preset work clothes detection model, and the parameters of the tracking algorithm are dynamically adjusted according to the actual location information and the parameters in the comparison table to ensure that the tracking accuracy is always in the optimal state.

[0128] In one embodiment, before step S1, the work clothes detection method combining yolov8 and target tracking further includes:

[0129] S10: Obtain the camera adjustment angle, and determine the recognition accuracy of the work clothes detection target according to the camera installation height and the camera adjustment angle.

[0130] In this embodiment, the camera adjustment angle refers to the angle of the camera relative to the horizontal plane; the camera adjustment angle is read from the camera configuration file or metadata. Recognition accuracy refers to the camera's ability to recognize work clothes detection targets at a specific angle.

[0131] Specifically, based on the camera installation height and camera adjustment angle, the effective recognition area within the camera field of view is calculated, and the recognition accuracy at different adjustment angles is verified through experimental data and historical records.

[0132] S20: Obtaining a first coefficient for adjusting the first predicted position according to the recognition accuracy of the work clothes detection target.

[0133] Specifically, the first coefficient is calculated based on the recognition accuracy. The higher the recognition accuracy, the closer the first coefficient is to 1; the lower the recognition accuracy, the larger the first coefficient is, so as to compensate for the recognition error.

[0134] In this embodiment, the first predicted position of the work clothes detection target when entering the monitoring area is determined based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, specifically including:

[0135] S100: Determine a first predicted position of the work clothes detection target when it enters the monitoring area based on the first coefficient, the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target.

[0136] Specifically, the predicted position is adjusted according to the first coefficient to improve the accuracy of the prediction.

[0137] It should be understood that the serial numbers of the steps in the above embodiments do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0138] In one embodiment, a work clothes detection system combining yolov8 and target tracking is provided. The work clothes detection system combining yolov8 and target tracking corresponds to a work clothes detection method combining yolov8 and target tracking in the above embodiment.

[0139] A work clothes detection system combining YOLOv8 and target tracking, including modules A, B, C and D. The detailed description of each functional module is as follows:

[0140] Data acquisition module, used to obtain surveillance video data of the target surveillance area and perform data preprocessing;

[0141] The work clothes detection module is used to perform work clothes target detection on the pre-processed surveillance video data in the preset work clothes detection model to obtain the work clothes detection results; the preset work clothes detection model is constructed by combining the YOLOv8 algorithm and the target tracking algorithm;

[0142] The target tracking module is used to perform target tracking analysis based on the work clothes detection results and obtain the work clothes tracking results of the continuous video stream; the defect detection module is used to generate defect detection marks based on the work clothes detection results and trigger alarm instructions;

[0143] The geographic location module is used to obtain geographic location information based on the work uniform tracking results, perform spatial matching analysis based on the geographic location information and the work uniform tracking results, and obtain the safety target detection results of the corresponding target monitoring area.

[0144] For the specific limitations of the work clothes detection system combined with yolov8 and target tracking, please refer to the limitations of the work clothes detection method combined with yolov8 and target tracking in the above text, which will not be repeated here; the various modules in the above-mentioned work clothes detection system combined with yolov8 and target tracking can be implemented in whole or in part through software, hardware and their combination; the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0145] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store monitoring video data and work clothes detection models, etc. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a work clothes detection method combining yolov8 and target tracking is implemented.

[0146] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are performed:

[0147] S1: Acquire surveillance video data of the target surveillance area and perform data preprocessing;

[0148] S2: In the preset work clothes detection model, work clothes target detection is performed on the pre-processed surveillance video data to obtain the work clothes detection results; based on the work clothes detection results, target tracking analysis is performed to obtain the work clothes tracking results of the continuous video stream; the preset work clothes detection model is constructed by combining the YOLOv8 algorithm and the target tracking algorithm;

[0149] S3: Generate defect detection marks based on the work clothes inspection results and trigger alarm instructions;

[0150] S4: Obtain geographic location information based on the work uniform tracking results, perform spatial matching analysis based on the geographic location information and the work uniform tracking results, and obtain the safety target detection results of the corresponding target monitoring area.

[0151] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0152] S1: Acquire surveillance video data of the target surveillance area and perform data preprocessing;

[0153] S2: In the preset work clothes detection model, work clothes target detection is performed on the pre-processed surveillance video data to obtain the work clothes detection results; based on the work clothes detection results, target tracking analysis is performed to obtain the work clothes tracking results of the continuous video stream; the preset work clothes detection model is constructed by combining the YOLOv8 algorithm and the target tracking algorithm;

[0154] S3: Generate defect detection marks based on the work clothes inspection results and trigger alarm instructions;

[0155] S4: Obtain geographic location information based on the work uniform tracking results, perform spatial matching analysis based on the geographic location information and the work uniform tracking results, and obtain the safety target detection results of the corresponding target monitoring area.

[0156] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0157] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0158] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A work clothes detection method combining yolov8 and target tracking, characterized in that: include: Obtain surveillance video data of the target surveillance area and perform data preprocessing; In the preset work clothes detection model, the pre-processed surveillance video data is subjected to work clothes target detection to obtain the work clothes detection results; Performing target tracking analysis based on the work clothes detection results to obtain work clothes tracking results of a continuous video stream; The preset work clothes detection model is constructed by combining the yolov8 algorithm and the target tracking algorithm; generating a defect detection mark based on the work clothes inspection result and triggering an alarm instruction; Acquiring geographic location information based on the work uniform tracking result, performing spatial matching analysis based on the geographic location information and the work uniform tracking result to obtain a safety target detection result corresponding to the target monitoring area; After acquiring the surveillance video data of the target surveillance area and performing data preprocessing, the method further includes: Determine, based on the monitoring video data, a first predicted position of the work clothes detection target when it enters the monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area; generating a detection target tracking comparison table for adjusting the accuracy of work clothes detection target tracking based on the monitoring video data, the first predicted position, and the second predicted position; Obtaining a target position change reference range representing a target position change threshold for work clothes detection, and performing adaptive parameter optimization on the target tracking algorithm based on the target position change reference range and the detection target tracking comparison table; Acquiring position information of a first actual work clothes detection target, and inputting the position information of the first actual work clothes detection target into the preset work clothes detection model to adjust the tracking accuracy of the work clothes detection target; The monitoring video data includes the camera type, camera installation height, distance between the camera and the monitoring area, and width and height of the monitoring area; Determining, based on the monitoring video data, a first predicted position of a work clothes detection target when it enters a monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area, specifically includes: Obtain parameter information of the work clothes detection model, including the detection accuracy of the yolov8 algorithm, the tracking accuracy of the target tracking algorithm, and the minimum and maximum sizes of the work clothes detection target; Determining a first predicted position of the work clothes detection target when it enters the monitoring area according to the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target; The second predicted position of the work clothes detection target when passing through a specific area is determined according to the camera type, the width and height of the monitoring area, the maximum size of the work clothes detection target, and the tracking accuracy of the target tracking algorithm.

2. A work clothes detection method combining yolov8 and target tracking according to claim 1, characterized in that, The acquisition of surveillance video data of the target surveillance area and data preprocessing specifically includes: Performing frame processing on the surveillance video data of the target surveillance area to obtain multiple frames of continuous surveillance images to be determined; In the preset work clothes detection model, work clothes target detection is performed on the pre-processed monitoring video data to obtain a work clothes detection result; target tracking analysis is performed based on the work clothes detection result to obtain a work clothes tracking result of the continuous video stream, including: In the preset work clothes detection model, work clothes target detection is performed frame by frame on multiple frames of the monitoring images to be judged based on the yolov8 algorithm to obtain the initial work clothes detection results of the corresponding continuous multiple frames of monitoring images; the initial work clothes detection results include several work clothes detection targets; In a preset work clothes detection model, target tracking analysis is performed on several work clothes detection targets based on the target tracking algorithm and the initial work clothes detection result to obtain a work clothes tracking result of a continuous video stream.

3. A work clothes detection method combining yolov8 and target tracking according to claim 2, characterized in that, The acquiring of geographic location information based on the work uniform tracking result, and performing spatial matching analysis based on the geographic location information and the work uniform tracking result to obtain a safety target detection result corresponding to the target monitoring area specifically include: Obtaining the geographical location coordinates of the detection device of the monitoring video data, and performing plane coordinate projection conversion on the geographical location coordinates to obtain plane rectangular coordinates; Based on the plane rectangular coordinates, when performing frame processing on the monitoring video data, the image coordinates of each frame of the monitoring image are positioned using a linear interpolation method to obtain frame image positioning information; Determining the geographic location information of the monitoring image of the work clothes tracking result based on the plane rectangular coordinate system and the frame image positioning information; Tracking monitoring is based on the geographic location information of the same work clothes detection target in multiple work clothes tracking results of continuous video streams to obtain the safety target detection results of the corresponding target monitoring area.

4. A work clothes detection method combining yolov8 and target tracking according to claim 1, characterized in that, Generating a detection target tracking comparison table for adjusting the work clothes detection target tracking accuracy based on the monitoring video data, the first predicted position, and the second predicted position specifically includes: Obtaining a predicted trajectory length of the work clothes detection target based on the width and height of the monitoring area, the first predicted position, and the second predicted position; According to the detection accuracy of the yolov8 algorithm and the tracking accuracy of the target tracking algorithm, the predicted detection time and the predicted tracking time of the work clothes detection target are obtained; Obtaining a predicted tracking speed of the work clothes detection target according to the predicted trajectory length of the work clothes detection target, the predicted detection time, and the predicted tracking time; Calculating a displacement distance between the first predicted position and the second predicted position to obtain a predicted displacement ratio for predicting an actual displacement distance; Calculate the detection and tracking time difference between the predicted detection time and the predicted tracking time, and calculate the tracking speed difference between the predicted tracking speed and the actual tracking speed of the work clothes detection target; Obtaining a predicted tracking time correspondence for determining an actual tracking speed of a work clothes detection target based on the detection tracking time difference and the tracking speed difference; According to the predicted displacement ratio and the predicted tracking speed-time correspondence, a detection target tracking comparison table for comparing and adjusting the tracking accuracy of the work clothes detection target is generated.

5. A work clothes detection method combining yolov8 and target tracking according to claim 1, characterized in that, Before determining the first predicted position of the work clothes detection target when it enters the monitoring area based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, the method further includes: Obtaining a camera adjustment angle, and determining recognition accuracy of a work clothes detection target based on the camera installation height and the camera adjustment angle; Obtaining a first coefficient for adjusting the first predicted position according to the recognition accuracy of the work clothes detection target; The determining, based on the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target, of a first predicted position when the work clothes detection target enters the monitoring area specifically includes: A first predicted position of the work clothes detection target entering the monitoring area is determined based on the first coefficient, the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target.

6. A work clothes detection system combining yolov8 and target tracking, characterized in that: A method for detecting work clothes in combination with Yolov8 and target tracking as described in any one of claims 1 to 5, wherein the system comprises: Data acquisition module, used to obtain surveillance video data of the target surveillance area and perform data preprocessing; A work clothes detection module is used to perform work clothes target detection on the pre-processed monitoring video data in a preset work clothes detection model to obtain a work clothes detection result; the preset work clothes detection model is constructed by combining the YOLOv8 algorithm and the target tracking algorithm; A target tracking module, configured to perform target tracking analysis based on the work clothes detection results to obtain work clothes tracking results of a continuous video stream; A defect detection module, configured to generate a defect detection mark based on the work clothes detection result and trigger an alarm instruction; A geographic location module is used to obtain geographic location information based on the work uniform tracking result, perform spatial matching analysis based on the geographic location information and the work uniform tracking result, and obtain a safety target detection result corresponding to the target monitoring area; After acquiring the surveillance video data of the target surveillance area and performing data preprocessing, the method further includes: Determine, based on the monitoring video data, a first predicted position of the work clothes detection target when it enters the monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area; generating a detection target tracking comparison table for adjusting the accuracy of work clothes detection target tracking based on the monitoring video data, the first predicted position, and the second predicted position; Obtaining a target position change reference range representing a target position change threshold for work clothes detection, and performing adaptive parameter optimization on the target tracking algorithm based on the target position change reference range and the detection target tracking comparison table; Acquiring position information of a first actual work clothes detection target, and inputting the position information of the first actual work clothes detection target into the preset work clothes detection model to adjust the tracking accuracy of the work clothes detection target; The monitoring video data includes the camera type, camera installation height, distance between the camera and the monitoring area, and width and height of the monitoring area; Determining, based on the monitoring video data, a first predicted position of a work clothes detection target when it enters a monitoring area and a second predicted position of the work clothes detection target when it passes through a specific area, specifically includes: Obtain parameter information of the work clothes detection model, including the detection accuracy of the yolov8 algorithm, the tracking accuracy of the target tracking algorithm, and the minimum and maximum sizes of the work clothes detection target; Determining a first predicted position of the work clothes detection target when it enters the monitoring area according to the camera installation height, the distance between the camera and the monitoring area, and the minimum size of the work clothes detection target; The second predicted position of the work clothes detection target when passing through a specific area is determined according to the camera type, the width and height of the monitoring area, the maximum size of the work clothes detection target, and the tracking accuracy of the target tracking algorithm.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the work clothes detection method combining yolov8 and target tracking as described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a work clothes detection method combining yolov8 and target tracking are implemented as described in any one of claims 1 to 5.

Images