A safety rope wearing state monitoring method and system

Abstract

The application provides a safety rope wearing state monitoring method, comprising the following steps: step A: when a vehicle appears in a monitoring area, monitoring a roof state through visual analysis, saving an image when a person is identified on the roof, and recording a portrait area coordinate; step B: finding a safety rope position in the image saved in step A through visual analysis, and recording a safety rope coordinate; step C: taking an intersection of the portrait area coordinate and the safety rope coordinate, and if the intersection is not empty, regarding that a worker has worn the safety rope, and if the intersection is empty, regarding that the worker has not worn the safety rope. The application also provides a monitoring system corresponding to the method. The application has the advantages that: the safety rope position is found in the picture in which the portrait is monitored, and whether the safety rope is worn is judged through the coordinate intersection, the operation amount is small, the processing speed is fast, the judgment logic is simple and effective, the result can be given in real time, the control center can be fed back in time, and the monitoring on the safety rope wearing state of the worker is effectively realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of visual analysis technology, in particular to a safety rope wearing state monitoring method and system. BACKGROUND

[0002] Safety rope is a protective device that cannot be avoided in high-altitude operation. The rate of injury and death accidents caused by falling of personnel without wearing safety ropes is quite high in industrial production engineering, and most of them are fatal accidents. Therefore, the safety rope for high-altitude operation is the lifeline of the staff. However, in the process of cement transportation and delivery, the driver needs to climb onto the roof to wear a safety rope before operation. Since the transportation driver is generally not a factory employee, there is no good management method, and he often does not operate according to the requirements, which poses a great hidden danger to safety production. At present, the wearing state of the safety rope is monitored by manual inspection, and there is no good method to realize unmanned and intelligent detection. At present, the industry mainly uses CN102512773B based on a gyroscope acceleration sensor or CN208893510U to add a pressure, infrared sensor and other devices on the body part to realize remote monitoring of the safety rope wearing state. This method needs to add new devices, and the wearing method of the monitoring object is more complex, which is not convenient to use, and may further lead to the unwillingness of the operator to wear the safety rope, and has certain limitations in popularization. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a safety rope wearing state monitoring method based on visual analysis. The present application solves the above technical problems by the following technical scheme: a safety rope wearing state monitoring method, comprising

[0004] Step A: when a vehicle appears in the monitoring area, monitor the roof state through visual analysis, save the image when a person is identified on the roof, and record the portrait area coordinates;

[0005] Step B: find the position of the safety rope in the image saved in step A through visual analysis, and record the coordinates of the safety rope;

[0006] Step C: take the intersection of the portrait area and the safety rope coordinates, if the intersection is not empty, it is considered that the operator has worn the safety rope, and if the intersection is empty, the safety rope is not worn.

[0007] The present application finds the position of the safety rope in the picture where the portrait is monitored, judges whether the safety rope is worn through the coordinate intersection, has small calculation amount, fast processing speed, simple and effective judgment logic, can give real-time results, feedbacks the control center in time, and effectively realizes the monitoring of the safety rope wearing state of the operator.

[0008] Preferably, the method for determining the presence of a vehicle in the monitoring area in step A is to set a photoelectric switch in the working area of the vehicle, and the vehicle entering the working area triggers the photoelectric switch.

[0009] Preferably, after the photoelectric switch is triggered, the method further comprises the steps of identifying and storing the license plate information.

[0010] Preferably, a camera is arranged in the monitoring area, a roof personnel detection algorithm is called to process the pictures obtained by the camera, when a person on the roof is detected, the portrait area coordinates are recorded, and the image is transmitted to the safety rope detection algorithm.

[0011] Preferably, the roof personnel detection algorithm is obtained by collecting image data and learning using a yolo network algorithm, wherein the image data includes multi-scene image data of no vehicle and no person, vehicle and person, vehicle and no person, and vehicle and person in the monitoring area, the vehicle and person scene includes a person on the roof and a person beside the vehicle, and the person on the roof scene is labeled as the same category, and other scenes are labeled as another category, and the image data is randomly scaled, rotated, and contrast adjusted, and the labeled and operated image is input into the yolo network algorithm to learn the roof personnel detection algorithm.

[0012] Preferably, the safety rope detection algorithm is obtained by collecting image data and learning using a yolo network algorithm, and the image data includes images of three states of no safety rope, safety rope natural suspension, and safety rope connected with human body, the safety rope connected with human body is labeled as normal state, and other images are labeled as abnormal state, and the labeled image is input into the yolo network algorithm to learn the safety rope detection algorithm.

[0013] Preferably, a plurality of cameras with different viewing angles are arranged in the monitoring area, and pictures of the roof personnel and the safety rope in different states are collected for each viewing angle camera for training, and the pictures are respectively acquired for identification.

[0014] Preferably, the roof personnel detection algorithm obtains the pictures from the camera in a polling manner, detects once per second, until a person on the roof signal is detected, saves the portrait area coordinates, and sends the picture to the safety rope detection algorithm, detects the area coordinates of the safety rope, takes the intersection of the portrait area and the safety rope area, and when the intersection is not empty, it is considered that the safety rope has been normally worn, and when the intersection is empty, the detection continues, if no signal of wearing the safety rope is detected within a preset time period after detecting the person on the roof, an alarm signal is sent to the control center, and the video is saved.

[0015] The application also provides a safety rope wearing state monitoring system, comprising

[0016] A camera is arranged in the monitoring area, a roof personnel detection algorithm is called to process the pictures obtained by the camera, when a person on the roof is detected, the portrait area coordinates are recorded, and the image is transmitted to the safety rope detection algorithm.

[0017] A license plate recognition module is configured to detect whether there is a vehicle in the monitoring area and obtain license plate information of the vehicle.

[0018] A roof personnel detection module is configured to, when a vehicle appears in the monitoring area, analyze the monitoring roof state through vision, save an image when a person is identified on the roof, and record a portrait area coordinate.

[0019] A safety rope detection module is configured to analyze the image saved in step A through vision to find a position of the safety rope and record a coordinate of the safety rope.

[0020] A judgment module is configured to take an intersection of the portrait area and the safety rope coordinate, and if the intersection is not empty, it is considered that the working person has worn the safety rope, and if the intersection is empty, it is considered that the safety rope is not worn.

[0021] Preferably, the method for judging whether a vehicle appears in the monitoring area comprises the following steps: setting an optical switch in a vehicle working area, triggering the optical switch when the vehicle enters the working area; and after the optical switch is triggered, recognizing and storing the license plate information.

[0022] The roof personnel detection module and the safety rope detection module respectively collect pictures of personnel and safety ropes in different states, manually label, learn through a yolo network algorithm after the labeling is completed, and complete training.

[0023] The roof personnel detection module obtains pictures from a camera through polling, detects once per second, detects a person-on-roof signal, saves a portrait area coordinate, sends the picture to the safety rope detection module, detects a safety rope area coordinate, takes an intersection of the portrait area and the safety rope area, considers that the safety rope has been normally worn when the intersection is not empty, continues to detect when the intersection is empty, sends an alarm signal to a control center and saves a video if a safety rope wearing signal is not detected within a preset time period after the person-on-roof signal is detected.

[0024] The safety rope wearing state monitoring method and system have the following advantages: the safety rope position is found in the picture of the monitored portrait, whether the safety rope is worn is judged through the coordinate intersection, the operation amount is small, the processing speed is fast, the judgment logic is simple and effective, the result can be given in real time, the control center is fed back in time, and the monitoring of the safety rope wearing state of the working personnel is effectively realized. The recognition model is obtained by manually labeling the pictures of different scenes, the recognition accuracy is high, and the recognition accuracy will be higher and higher with the increase of the use feedback, the yolo algorithm can detect the object state and the area in the image, the intersection operation is facilitated, the analysis is performed through multiple perspectives, and the result is ensured to be accurate. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 A flowchart of the safety rope wearing state monitoring method provided for the embodiments of the application is shown. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] like Figure 1 As shown, this embodiment provides a method for monitoring the wearing status of a safety rope, including...

[0028] Step A: When a vehicle appears in the monitoring area, monitor the status of the vehicle roof through visual analysis. When a person is identified on the vehicle roof, save the image and record the coordinates of the person's image area.

[0029] Step B: Locate the safety rope in the image saved in Step A using visual analysis, and record the coordinates of the safety rope;

[0030] Step C: Find the intersection of the coordinates of the human figure area and the safety rope. If the intersection is not empty, it is assumed that the worker is wearing a safety rope. If the intersection is empty, the worker is not wearing a safety rope.

[0031] This embodiment locates the safety rope in the image of the monitored person and determines whether the safety rope is being worn by using the intersection of coordinates. It has low computational load, fast processing speed, simple and effective judgment logic, and can provide results in real time and timely feedback to the control center, effectively realizing the monitoring of the safety rope wearing status of the workers.

[0032] Specifically, the method for determining the presence of a vehicle within the monitoring area is as follows: a photoelectric switch is set up in the vehicle's working area, and the photoelectric switch is triggered when the vehicle enters the working area.

[0033] The monitoring area is also equipped with cameras. When the photoelectric switch is triggered, the cameras can identify and store license plate information for later viewing of vehicle status.

[0034] When detecting people on the roof, the roof-mounted personnel detection algorithm detects images obtained by the camera. When a person is detected on the roof, the coordinates of the person's image area are recorded, and the image is transmitted to the safety rope detection algorithm.

[0035] The roof personnel detection algorithm is obtained by collecting image data and learning by using a YOLO network algorithm, wherein the image data includes multi-scene image data of no car and no person, car and person, car and no person, and no car and person in a monitoring area, the car and person scene includes a case of a person on the roof of the car and a person beside the car, and the case of a person on the roof of the car is labeled as the same category, and other scenes are labeled as another category, and the image data is randomly scaled, rotated (less than 30 degrees), and contrast image adjusted, and the labeled and operated image is input into the YOLO network algorithm to learn the roof personnel detection algorithm.

[0036] The data and labels are learned by the YOLO network algorithm, and a network model is output. When detecting, an image is input, and the model can be used to predict the result of roof detection and the image position area of a person.

[0037] The safety rope detection algorithm is obtained by collecting image data and learning by using a YOLO network algorithm, and the image data includes images of three states of no safety rope, natural suspension of the safety rope, and connection of the safety rope and the human body, the connection of the safety rope and the human body is labeled as a normal state, and other images are labeled as an abnormal state, and the labeled image is input into the YOLO network algorithm to learn the safety rope detection algorithm.

[0038] To ensure the accuracy of the detection result, a plurality of cameras with different viewing angles are arranged in the monitoring area in the preferred embodiment, the influence of the overlapping of the viewing angles on the detection result is eliminated by multi-view detection, when the safety rope is detected by part of the viewing angles, verification is performed by other viewing angles, and only when the monitoring results of all the cameras with different viewing angles indicate that the worker has worn the safety rope, it is considered that the worker has worn the safety rope and can normally work.

[0039] Preferably, the safety rope also needs to be connected with a fixed rod, if necessary, the positions of the safety rope and the fixed rod can be detected at the same time, and it is considered that the safety rope has been connected with the fixed rod when the coordinates of the safety rope and the coordinates of the fixed rod have an intersection, so as to prevent accidents caused by the safety rope not being fixed.

[0040] The roof personnel detection algorithm obtains pictures from the camera by polling, detects once per second until a person on the roof signal is detected, saves the portrait area coordinates, and sends the picture to the safety rope detection algorithm to detect the area coordinates of the safety rope, takes the intersection of the portrait area and the safety rope area, considers that the safety rope has been normally worn when the intersection is not empty, and continues to detect when the intersection is empty, if no signal of wearing the safety rope is detected within a preset time (30 seconds in this embodiment) after detecting a person on the roof, an alarm signal is sent to the control center, and a video is saved.

[0041] The embodiment supports simultaneous addition of multiple cameras for multi-angle safety rope detection to avoid or reduce the missed detection problem caused by shielding. The camera polling detection method is used to detect whether there is a person on the roof. Pictures are obtained from the camera and sent to the roof personnel detection algorithm. Detection is performed once every second until a person is detected on the roof. The position area information A is saved. The picture is sent to the safety rope detection algorithm to detect the safety rope position information B. The region intersection ratio IoU of the working person and the safety rope is calculated. If the IoU is greater than 0, it is judged that the person wears a safety rope for work. Otherwise, it indicates that the person does not wear a safety rope for work. If no safety rope is detected, the camera polling stage is returned to continue detection.

[0042] The system detection decision adopts multiple cameras to realize multi-angle safety rope wearing detection analysis and recognition. If no safety rope is detected by multiple cameras within a period of time, the central system is warned to remind the user that there may be a safety rope wearing behavior. At the same time, suspicious violation videos are saved, and license plate information is recorded.

[0043] The embodiment also provides a safety rope wearing state monitoring system, which comprises

[0044] A camera is configured to obtain a picture of a monitoring area.

[0045] A license plate recognition module is configured to detect whether there is a vehicle in the monitoring area and obtain license plate information of the vehicle.

[0046] A roof personnel detection module is configured to analyze the monitoring area by vision when a vehicle appears in the monitoring area, save an image when a person is identified on the roof, and record the coordinates of the image area.

[0047] A safety rope detection module is configured to find the position of the safety rope in the image saved in step A by visual analysis and record the coordinates of the safety rope.

[0048] A judgment module is configured to take the intersection of the coordinates of the image area and the safety rope. If the intersection is not empty, it is considered that the working person has worn a safety rope. If the intersection is empty, it is considered that the working person has not worn a safety rope.

[0049] The method for judging whether a vehicle appears in the monitoring area comprises the following steps: setting an optical switch in a vehicle working area; triggering the optical switch when the vehicle enters the working area; and identifying and storing the license plate information after the optical switch is triggered.

[0050] The roof personnel detection module and the safety rope detection module respectively collect pictures of personnel and safety ropes in different states, and perform manual labeling. After the labeling is completed, the yolo network algorithm is used for learning and training.

[0051] The roof person detection module obtains pictures from the camera by polling, detects once per second, until a roof person signal is detected, saves the person image area coordinates, and sends the picture to the safety harness detection module, detects the safety harness area coordinates, takes the intersection of the person image area and the safety harness area, when the intersection is not empty, it is considered that the safety harness is normally worn, when the intersection is empty, it continues to detect, if after detecting the roof person, the signal of wearing the safety harness is not detected within the preset time, an alarm signal is sent to the control center, and the video is saved.

[0052] The above examples are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A safety rope wear state monitoring method characterized by: Comprising Step A: a plurality of cameras with different viewing angles are arranged in the monitoring area, and a photoelectric switch is arranged in the working area of the vehicle. When the vehicle enters the working area and triggers the photoelectric switch, the roof state is monitored through visual analysis, the camera calls the roof personnel detection algorithm to process the picture obtained by the camera, and when a person is identified on the roof, the image is saved, the portrait area coordinates are recorded, and the image is transmitted to the safety rope detection algorithm; The camera of each viewing angle respectively collects pictures of the roof personnel and the safety rope in different states for training, and respectively obtains pictures for identification; the roof personnel detection algorithm obtains pictures from the camera through polling, detects once per second, until a roof person signal is detected, saves the portrait area coordinates, and sends the picture to the safety rope detection algorithm; Step B: find the position of the safety rope in the image saved in step A through visual analysis, and record the coordinates of the safety rope; Step C: take the intersection of the coordinates of the portrait area and the safety rope. If the intersection is not empty, it is considered that the worker has worn the safety rope. If the intersection is empty, the safety rope is not worn. At this time, continue to detect. If no signal of wearing the safety rope is detected within a preset time after detecting a person on the roof, an alarm signal is sent to the control center, and the video is saved.

2. The method of claim 1, wherein: After the photoelectric switch is triggered, the step of identifying and storing the license plate information is further included.

3. The method of claim 1, wherein: The roof personnel detection algorithm is obtained by learning using a yolo network algorithm based on image data, wherein the image data includes multi-scene image data of no car and no person, car and person, car and no person, and no car and person in the monitoring area, the car and person scene includes a person on the roof and a person beside the car, and the person on the roof scene is labeled as the same category, and other scenes are labeled as another category. At the same time, the image data is randomly scaled, rotated, and contrast adjusted. The labeled and operated image is input into the yolo network algorithm to learn the roof personnel detection algorithm.

4. The method of claim 1, wherein: The safety rope detection algorithm is obtained by learning using a yolo network algorithm based on image data, wherein the image data includes images of three states of no safety rope, natural suspension of safety rope, and connection of safety rope with human body. The connection of safety rope with human body is labeled as a normal state, and other images are labeled as an abnormal state. The labeled image is input into the yolo network algorithm to learn the safety rope detection algorithm.

5. A safety line wear state monitoring system characterized by: Comprising A camera, a plurality of cameras with different viewing angles are arranged in the monitoring area, for obtaining the picture of the monitoring area; A photoelectric switch is arranged in the working area of the vehicle. When the vehicle enters the working area and triggers the photoelectric switch, the roof state is monitored through visual analysis, the camera calls the roof personnel detection algorithm to process the picture obtained by the camera, and when a person is identified on the roof, the image is saved, the portrait area coordinates are recorded, and the image is transmitted to the safety rope detection algorithm; The camera of each viewing angle respectively collects pictures of the roof personnel and the safety rope in different states for training, and respectively obtains pictures for identification; the roof personnel detection algorithm obtains pictures from the camera through polling, detects once per second, until a roof person signal is detected, saves the portrait area coordinates, and sends the picture to the safety rope detection algorithm; The safety rope detection module finds the position of the safety rope in the image saved in the camera through visual analysis and records the coordinates of the safety rope. The judgment module takes the intersection of the coordinates of the human image region and the safety rope, and if the intersection is not empty, it is considered that the worker has worn the safety rope, and if the intersection is empty, it is considered that the worker has not worn the safety rope, at which time the detection is continued.

6. A safety rope wear state monitoring system according to claim 5, characterized in that: The method for judging the presence of a vehicle in the monitoring area is to set an optical switch in the vehicle working area, and the vehicle enters the working area to trigger the optical switch; after the optical switch is triggered, the steps of identifying and storing the license plate information are further included; The roof personnel detection module and the safety rope detection module respectively collect pictures of personnel and safety ropes in different states and perform manual labeling, and after the labeling is completed, the training is completed through the yolo network algorithm; The roof personnel detection module obtains pictures from the camera through polling, detects once per second, until the roof person signal is detected, saves the human image region coordinates, and sends the pictures to the safety rope detection module to detect the safety rope region coordinates, takes the intersection of the human image region and the safety rope region, and when the intersection is not empty, it is considered that the safety rope has been normally worn, and when the intersection is empty, the detection is continued, if after detecting the roof person, the signal of wearing the safety rope is not detected within a preset time, an alarm signal is sent to the control center, and the video is saved.

Images