Limited space operation supervision method and system assisting AI vision

Abstract

The invention provides a limited space operation supervision method and system for assisting AI vision, and relates to the technical field of anomaly detection, and the method comprises the steps: collecting an operation image through a high-definition camera, compressing and transcoding through a video server, and cutting and detecting through an industrial control computer to generate operation region image data; and the detection model detects the image data to obtain three detection results, risk assessment is performed according to the results, a control instruction is generated, an alarm is controlled, and real-time monitoring and safety management of limited space operation are realized. The method can solve the problems that in the prior art, real-time safety detection on a working site is difficult to achieve, detection cannot be conducted on a specific area, the detection efficiency is low, only a working area is used for detection, the detection accuracy is improved, false alarms are reduced, intelligent judgment on potential risks can be achieved, and the safety of the working site is improved. Automatic alarm and intervention on a working site are realized, and the working safety is improved.

Description

Technical Field

[0001] The present application relates to the field of anomaly detection technology, and in particular to a method and system for monitoring confined space operations assisted by AI vision. Background Art

[0002] With the continuous advancement of industrialization, the safety of the work site has become an important issue in the process of enterprise development. In modern industrial production, especially in some harsh environments with limited work sites, such as in the chemical industry, if the work site is limited and the work environment is not ventilated and the harmful gas content is not detected, the harmful gas at the work site will exceed the standard, which will cause the workers to be further exposed to the dangerous working environment. Various harmful substances in the environment will cause a series of health problems for the workers, such as respiratory diseases, skin diseases, poisoning, etc. Due to its special environmental conditions, confined space operations have many potential dangers, so the requirements for work safety are very high. Traditional manual supervision methods are difficult to achieve comprehensive and real-time monitoring.

[0003] At present, existing technologies make it difficult to achieve real-time detection of all areas of the work site and cannot detect specific areas, resulting in low detection efficiency. Summary of the invention

[0004] The purpose of this application is to provide a limited space operation supervision method and system assisted by AI vision, so as to solve the technical problem that the existing technology is difficult to achieve real-time detection of all areas of the work site and cannot detect specific areas, resulting in low detection efficiency.

[0005] In view of the above problems, the present application provides a limited space operation supervision method and system assisted by AI vision.

[0006] In the first aspect, the present application provides a limited space operation supervision method assisted by AI vision, and the limited space operation supervision method assisted by AI vision is implemented by a limited space operation supervision system assisted by AI vision. Among them, the limited space operation supervision method assisted by AI vision includes: using the image acquisition device to collect high-definition operation image data of the target area in real time; compressing and transcoding the high-definition operation image data through the video server to generate packaged data; inputting the packaged data into the industrial computer, the industrial computer includes a cropping unit and a detection unit, and the cropping unit performs ROI cropping on the packaged data according to the operation area to generate operation area image data; the operation area image data is detected by the operation detection model deployed in the detection unit to obtain the operation detection result, including the first detection result, the second detection result, and the third detection result; combining the first detection result, the second detection result, and the third detection result to perform a risk assessment based on a preset rule to obtain a risk assessment result; generating a control instruction according to the risk assessment result, and controlling the alarm according to the control instruction.

[0007] In the second aspect, the present application also provides a limited space operation supervision system assisted by AI vision, which is used to execute the limited space operation supervision method assisted by AI vision as described in the first aspect. Among them, the limited space operation supervision system assisted by AI vision includes: a data acquisition module, which is used to use an image acquisition device to collect high-definition operation image data of the target area in real time; an image processing module, which is used to compress and transcode the high-definition operation image data through a video server to generate packaged data; an image cropping module, which is used to input the packaged data into an industrial computer, and the industrial computer includes a cropping unit and a detection unit, and the cropping unit performs ROI cropping on the packaged data according to the operation area to generate operation area image data; an image detection module, which is used to detect the operation area image data through the operation detection model deployed in the detection unit, and obtain the operation detection result, including the first detection result, the second detection result, and the third detection result; a risk assessment module, which is used to combine the first detection result, the second detection result, and the third detection result to perform risk assessment based on preset rules to obtain a risk assessment result; a control instruction generation module, which is used to generate a control instruction according to the risk assessment result, and control the alarm according to the control instruction.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] The high-definition operation image data of the target area is collected in real time by using the image acquisition device; the high-definition operation image data is compressed and transcoded by the video server to generate packaged data; the packaged data is input into the industrial computer, the industrial computer includes a cropping unit and a detection unit, the cropping unit performs ROI cropping on the packaged data according to the operation area to generate the operation area image data; the operation detection model deployed in the detection unit detects the operation area image data to obtain the operation detection result, including the first detection result, the second detection result, and the third detection result; the first detection result, the second detection result, and the third detection result are combined to perform a risk assessment based on a preset rule to obtain a risk assessment result; a control instruction is generated according to the risk assessment result, and the alarm is controlled according to the control instruction, which effectively solves the technical problem that the prior art is difficult to realize real-time detection of all areas of the operation site, cannot detect specific areas, resulting in low detection efficiency, so that only the operation area is used for detection, improves the accuracy of detection, reduces false alarms, can realize intelligent judgment of potential dangers, realize automatic alarm and intervention on the operation site, and improves operation safety.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 A flowchart of the AI ​​vision-assisted limited space operation supervision method for this application;

[0013] Figure 2 This is a schematic diagram of the structure of the limited space operation supervision system assisted by AI vision for this application.

[0014] Description of reference numerals:

[0015] Data acquisition module 11, image processing module 12, image cropping module 13, image detection module 14, risk assessment module 15, control instruction generation module 16. DETAILED DESCRIPTION

[0016] This application solves the technical problem that the existing technology is difficult to achieve real-time detection of all areas of the work site and cannot detect specific areas, resulting in low detection efficiency, by providing a limited space work supervision method and system assisted by AI vision. Only the work area is used for detection, which improves the accuracy of detection, reduces false alarms, and can realize intelligent judgment of potential dangers, realize automatic alarm and intervention on the work site, and improve work safety.

[0017] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0018] For example, please refer to the attached Figure 1 The present application provides a limited space operation supervision method assisted by AI vision, wherein the limited space operation supervision method assisted by AI vision is applied to a limited space operation supervision system assisted by AI vision, the limited space operation supervision system assisted by AI vision is communicatively connected with a safety specification detection system, the safety specification detection system includes an image acquisition device, a video server, an industrial computer, and an alarm, and the limited space operation supervision method assisted by AI vision specifically includes the following steps:

[0019] S1: Using the image acquisition device to acquire high-definition operation image data of the target area in real time.

[0020] Specifically, choose the right HD camera based on the size of the work environment, light conditions, type of work, and other factors. For example, in an underground mine with low light, an infrared camera is required. Install the camera in a position that can maximize coverage of the work area and minimize blind spots. Multiple cameras need to be installed at multiple angles and heights. Adjust the camera's resolution, frame rate, exposure time, and other parameters to ensure clear images even in dynamic or changing light conditions.

[0021] S2: compressing and transcoding the high-definition operation image data through the video server to generate packaged data.

[0022] Specifically, a video server is a device dedicated to processing and distributing video data. It is usually responsible for receiving the original video stream from the image acquisition device and processing it. Configure compression and transcoding parameters on the video server, including resolution, frame rate, bit rate, etc. Make sure that the server can handle video streams from multiple cameras. Compression is a technology that reduces the size of data by removing redundant information and reducing the required storage space and bandwidth. Video compression can be a variety of encoding standards, such as H.264, H.265 or VP9. Transcoding refers to the process of converting video from one encoding format to another. It is to adapt to different playback devices or network environments. For example, transcoding high-definition video into a format suitable for playback on mobile devices. Packaging data refers to encapsulating compressed and transcoded video data into a format suitable for transmission. Including encapsulating video streams into RTSP, RTMP or other streaming protocols for easy transmission on the network.

[0023] S3: inputting the packaged data into the industrial computer, wherein the industrial computer comprises a cutting unit and a detection unit, and the cutting unit performs ROI cutting on the packaged data according to the working area to generate working area image data.

[0024] Specifically, an industrial computer is an edge computing device designed for industrial environments. It has the characteristics of good stability and resistance to environmental interference, and is suitable for processing data on industrial sites. The cropping unit is a software or hardware module in the industrial computer. Its function is to crop a specific area, namely ROI, from the original image. ROI is the area in the image that the user is interested in or needs to focus on. ROI cropping refers to the process of extracting a specific area from the original image data. This can reduce the amount of calculation for subsequent image processing and improve processing speed and efficiency. The image data generated after ROI cropping only contains the operation area, which will be used for subsequent image analysis and operation detection. The cropping unit is configured on the industrial computer so that it can process the input image data according to the defined ROI. When the packaged data is input into the industrial computer, the cropping unit extracts the corresponding ROI from each frame of the image according to the preset parameters. The cropped image data will be saved as a new image file or data stream for use by the detection unit.

[0025] S4: Detecting the image data of the operation area through the operation detection model deployed in the detection unit to obtain operation detection results, including a first detection result, a second detection result, and a third detection result.

[0026] Specifically, the detection unit is a key component in the industrial computer, which is responsible for running the job detection model, processing the input image data, and outputting the detection results. The job detection model is a training model based on machine learning or deep learning algorithms, which can identify specific objects, behaviors or scenes in images. The first detection result, the second detection result, and the third detection result can be different types of analysis and recognition results. Among them, the first detection result is to detect whether there is a dedicated person holding a blower entering the working area, the second detection result is to detect whether there is a gas monitoring instrument in the working area; the third detection result is to detect whether there is a dedicated person in the working area to monitor the attributes and location information of three categories of targets;

[0027] S5: Perform risk assessment based on preset rules in combination with the first detection result, the second detection result, and the third detection result to obtain a risk assessment result.

[0028] Furthermore, the present application also includes:

[0029] If the first detection result, the second detection result and the third detection result cannot be obtained at the same time within a preset number of consecutive frames, a danger instruction is generated; at the same time, detection is continued, and if the first detection result, the second detection result and the third detection result can be obtained at the same time, the danger instruction is updated to a safety instruction.

[0030] Specifically, a threshold of the number of consecutive frames is preset by a technician in this field, for example, it may be set to 5 frames or 10 frames. This threshold indicates that within so many consecutive frames, it is expected that a specific object or event can be stably detected. The first, second and third detection results need to be obtained at the same time, and relevant objects or events, such as operators, safety equipment and equipment status, must be successfully detected in each frame. If three detection results cannot be obtained at the same time within the preset number of consecutive frames, a danger instruction will be generated. This instruction may be to activate an alarm, stop the operation or notify the manager to take action. If the system can obtain three detection results at the same time during the continuous detection process, it will update the previous danger instruction to a safety instruction. It means that the working environment is now safe and the operation can continue or the previous alarm state can be lifted.

[0031] S6: Generate a control instruction according to the risk assessment result, and control the alarm according to the control instruction.

[0032] Specifically, the hazard assessment results are a quantitative assessment of the safety status of the work site, including risk level, urgency or specific warning information. Control instructions are specific operational commands generated based on the hazard assessment results, which instruct the monitoring system on how to respond. These instructions can be to start an alarm, send a notification, suspend operations, etc. An alarm is an audible or visual device used to warn on-site personnel when a safety risk is detected. The alarm can be an on-site speaker, flashing light, SMS notification system or other communication device.

[0033] Furthermore, the image acquisition device also includes a high-definition camera, a WIFI transmitter, and a WIFI receiver. The high-definition camera is connected to the WIFI transmitter via PoE; the WIFI receiver is connected to the video server via a first preset interface; the video server is connected to the industrial computer via a second preset interface; and the industrial computer is connected to the alarm via a third preset interface.

[0034] Specifically, the HD camera is used to collect HD image data at the work site. It has the characteristics of high resolution, night vision function, wide-angle lens, etc. to ensure clear images under various lighting conditions. The HD camera is connected to the WIFI transmitter through PoE technology. PoE technology allows data and power to be transmitted simultaneously through Ethernet cables, simplifying installation and wiring. The WIFI receiver is used to receive image data from the WIFI transmitter. It is a wireless router or wireless access point that is responsible for converting wireless signals into wired signals for transmission to the video server. The WIFI receiver and the video server are connected by a first preset interface. This interface can be an Ethernet interface, an optical fiber interface, or other interface suitable for high-speed data transmission. The video server is responsible for receiving image data from the WIFI receiver, compressing and transcoding it, and then packaging the data to prepare for sending to the industrial computer. The video server and the industrial computer are connected by a second preset interface. This interface also needs to support high-speed data transmission, which can be USB, Ethernet, or other professional interfaces. The industrial computer includes a cropping unit and a detection unit, which are responsible for ROI cropping and job detection of the input image data, and then performing a hazard assessment based on the detection results. The industrial computer is connected to the alarm through a third preset interface. This interface can be a serial port, GPIO or other interface suitable for controlling the alarm. According to the control instructions generated by the industrial computer, the alarm will emit sound or light signals to warn on-site personnel and managers to pay attention to safety.

[0035] Furthermore, the present application also includes:

[0036] By deploying a high-definition camera at the top of the target area, the high-definition operation image data in the target area is collected along a preset collection direction, and the collected high-definition operation image data is transmitted to the WIFI transmitter; the WIFI transmitter transmits the collected high-definition operation image data to the WIFI receiver by wireless transmission; the WIFI receiver transmits the high-definition operation image data to the video server.

[0037] Specifically, the HD camera is deployed at the top of the target area, and the position and angle of the camera are adjusted as needed, and the collection direction is preset to ensure that all key parts of the operation area are covered. The camera monitors the target area in real time along the preset collection direction and collects HD operation image data. The WIFI transmitter receives image data from the HD camera. Since the camera is deployed in a place where it is difficult to wire, the use of the WIFI transmitter can easily realize wireless transmission of data. The WIFI transmitter transmits image data to the WIFI receiver via a wireless network. This wireless transmission method reduces the need for wiring and improves installation flexibility. The WIFI receiver is responsible for receiving the HD operation image data transmitted by the WIFI transmitter. The received image data is transmitted to the video server through the WIFI receiver. Including converting the wireless signal into a wired signal for the convenience of the video server to process. The video server receives the image data from the WIFI receiver, compresses and transcodes it, and generates packaged data suitable for transmission and storage.

[0038] Furthermore, step S3 of the present application also includes:

[0039] Convert the packaged data into HSV image data; create a mask based on the HSV image data based on a preset color detection range; fill the mask with a closing operation, and then open the filled data to obtain standard image data; calculate the contour area of ​​the standard image data, and take the largest contour area as the working area; crop the packaged data according to the working area to obtain the working area image data.

[0040] Specifically, RGB image data is used for display and storage, but in image analysis, the HSV color space is more capable of highlighting color information and facilitating image processing based on color features. Convert the RGB image data in the packaged data to HSV format. This can be achieved through functions in OpenCV or other image processing libraries. Create a mask based on the preset color detection interval. This interval defines the color range of interest, for example, you may only focus on objects of a certain color. The mask is a binary image used to highlight or isolate a specific color area in the image. Perform a closing operation on the mask to fill any small holes or gaps in the color area. Closing is a morphological operation that first dilates and then erodes, which can connect adjacent color blocks. Perform an opening operation on the filled data to smooth the edges of the color area and remove small noise points. The opening operation is the opposite of the closing operation, which first erodes and then dilates. The image data processed by the closing and opening operations is called the standard image data, which contains a clear working area. Calculate the area of ​​all contours in the standard image data and find the largest contour area. This largest contour represents the working area. Based on the calculated work area, the original packaged data is cropped to retain only the image data of the work area. Using the contour information of the work area, a ROI is created and the area is extracted from the original image.

[0041] Furthermore, step S4 of the present application also includes:

[0042] LabelImg is used to annotate the work area image data frame by frame; the annotated work area image data is divided into a training set and a validation set according to a preset ratio, wherein the training set is 1600 frames and the validation set is 400 frames; the work detection model is trained according to the training set to obtain detection weights; a performance evaluation is performed on the detection weights according to the validation set, and the optimal detection weights are screened out according to the evaluation results; a validation data set of a limited work space image data set is used to perform a performance evaluation on the obtained detection weights to quantify the weight detection accuracy and screen out the optimal detection weights; the work area image data is detected according to the TensorRT model, the optimal detection weights and the work detection model.

[0043] Specifically, LabelImg is a graphical annotation tool that allows users to add and edit bounding boxes for images and assign a label to each box. Open LabelImg, load the work area image data, and then draw bounding boxes for objects in the image frame by frame and mark the category of the object. For example, label workers, tools, equipment, etc. After the annotation is completed, LabelImg will generate an XML or JSON file containing the bounding box and label information of each image. These files will be used to train the job detection model. According to the preset ratio, the annotated work area image data is divided into a training set and a validation set. The training set is used for model learning, and the validation set is used to evaluate the performance of the model. In this embodiment, the training set contains 1600 frames and the validation set contains 400 frames. Use the training set to train the job detection model. This model may be a convolutional neural network or other deep learning model suitable for target detection. During the training process, the model will learn how to identify and locate objects in the image and generate detection weights, which are the core of the model and are used for subsequent detection tasks. Use the validation set to evaluate the performance of the trained detection weights. Evaluation indicators include precision, recall, F1 score, etc. Based on the evaluation results, the optimal detection weights with the best performance are screened out. These weights will be used for actual detection tasks. TensorRT is a deep learning inference engine that optimizes deep learning models to improve inference speed and efficiency. The optimal detection weights are integrated with the TensorRT model and the job detection model for real-time detection of job area image data.

[0044] Furthermore, step S6 of the present application also includes:

[0045] The harmful gas concentration in the target area is monitored in real time through the harmful gas sensor built into the wearable device. If the harmful gas concentration exceeds the preset concentration, a first alarm instruction is generated; the noise value in the target area is monitored in real time through the noise sensor. If the noise value exceeds the preset noise value, a second alarm instruction is generated; the control instruction is generated according to the first alarm instruction, the second alarm instruction and the risk assessment result.

[0046] Specifically, wearable harmful gas sensors, such as gas detectors or smart wearable devices, are deployed in the target area. These devices can be worn by the operators or fixed at specific locations in the operation area. The sensors monitor the concentration of harmful gases in the target area in real time, such as hydrogen sulfide, carbon monoxide, methane, etc. The system presets a safe concentration range for harmful gases. When the monitored concentration exceeds this range, the system triggers an alarm. If the concentration of harmful gases exceeds the preset concentration, the system generates a first alarm instruction and may take measures, such as stopping operations, evacuating personnel, notifying rescue, etc. Noise sensors, such as sound level meters or smart wearable devices, are deployed in the target area. These devices can be worn by the operators or fixed at specific locations in the operation area. The sensors monitor the noise value in the target area in real time. The system presets a safe level of noise. When the monitored noise value exceeds this level, the system triggers an alarm. If the noise value exceeds the preset noise value, the system generates a second alarm instruction and may take measures, such as reducing the noise source, reminding the operators to wear earplugs, notifying management personnel, etc. The system comprehensively evaluates the safety status of the operating environment based on the first alarm instruction, the second alarm instruction and the hazard assessment results. Based on the comprehensive evaluation results, control instructions are generated. These instructions may be to stop the operation, adjust the working conditions, remind personnel to take protective measures, etc.

[0047] In summary, the AI ​​vision-assisted limited space operation supervision method provided in this application has the following technical effects:

[0048] The high-definition operation image data of the target area is collected in real time by using the image acquisition device; the high-definition operation image data is compressed and transcoded by the video server to generate packaged data; the packaged data is input into the industrial computer, the industrial computer includes a cropping unit and a detection unit, the cropping unit performs ROI cropping on the packaged data according to the operation area to generate the operation area image data; the operation detection model deployed in the detection unit detects the operation area image data to obtain the operation detection result, including the first detection result, the second detection result, and the third detection result; the first detection result, the second detection result, and the third detection result are combined to perform a risk assessment based on preset rules to obtain a risk assessment result; a control instruction is generated according to the risk assessment result, and the alarm is controlled according to the control instruction, which effectively solves the technical problem that the existing technology is difficult to realize real-time detection of all areas of the operation site, cannot detect specific areas, and leads to low detection efficiency, so that only the operation area is used for detection, the detection accuracy is improved, and false alarms are reduced, and intelligent judgment of potential dangers can be realized, automatic alarm and intervention of the operation site can be realized, and operation safety is improved.

[0049] Embodiment 2: Based on the AI ​​vision-assisted limited space operation supervision method in the previous embodiment, the present application also provides an AI vision-assisted limited space operation supervision system, please refer to the attached Figure 2 , the AI ​​vision-assisted limited space operation supervision system includes:

[0050] The data acquisition module 11 is used to acquire high-definition operation image data of the target area in real time using an image acquisition device.

[0051] The image processing module 12 is used to compress and transcode the high-definition operation image data through a video server to generate packaged data.

[0052] The image cropping module 13 is used to input the packaged data into an industrial computer. The industrial computer includes a cropping unit and a detection unit. The cropping unit performs ROI cropping on the packaged data according to the operating area to generate operating area image data.

[0053] The image detection module 14 is used to detect the image data of the operation area through the operation detection model deployed in the detection unit to obtain operation detection results, including a first detection result, a second detection result, and a third detection result.

[0054] The risk assessment module 15 is used to perform risk assessment based on preset rules in combination with the first detection result, the second detection result, and the third detection result to obtain a risk assessment result.

[0055] The control instruction generating module 16 is used to generate a control instruction according to the risk assessment result, and control the alarm according to the control instruction.

[0056] Furthermore, the AI ​​vision-assisted limited space operation supervision system also includes a device connection module for:

[0057] The image acquisition device also includes a high-definition camera, a WIFI transmitter, and a WIFI receiver. The high-definition camera is connected to the WIFI transmitter via PoE; the WIFI receiver is connected to the video server via a first preset interface; the video server is connected to the industrial computer via a second preset interface; and the industrial computer is connected to the alarm via a third preset interface.

[0058] Furthermore, the AI ​​vision-assisted limited space operation supervision system also includes a data transmission module for:

[0059] By deploying a high-definition camera at the top of the target area, the high-definition operation image data in the target area is collected along a preset collection direction, and the collected high-definition operation image data is transmitted to the WIFI transmitter; the WIFI transmitter transmits the collected high-definition operation image data to the WIFI receiver by wireless transmission; the WIFI receiver transmits the high-definition operation image data to the video server.

[0060] Furthermore, the image cropping module 13 in the limited space operation supervision system assisted by AI vision is also used for:

[0061] Convert the packaged data into HSV image data; create a mask based on the HSV image data based on a preset color detection range; fill the mask with a closing operation, and then open the filled data to obtain standard image data; calculate the contour area of ​​the standard image data, and take the largest contour area as the working area; crop the packaged data according to the working area to obtain the working area image data.

[0062] Furthermore, the image detection module 14 in the limited space operation supervision system assisted by AI vision is also used for:

[0063] LabelImg is used to annotate the work area image data frame by frame; the annotated work area image data is divided into a training set and a validation set according to a preset ratio, wherein the training set is 1600 frames and the validation set is 400 frames; the work detection model is trained according to the training set to obtain detection weights; a performance evaluation is performed on the detection weights according to the validation set, and the optimal detection weights are screened out according to the evaluation results; a validation data set of a limited work space image data set is used to perform a performance evaluation on the obtained detection weights to quantify the weight detection accuracy and screen out the optimal detection weights; the work area image data is detected according to the TensorRT model, the optimal detection weights and the work detection model.

[0064] Furthermore, the hazard assessment module 15 in the AI ​​vision-assisted limited space operation supervision system is also used for:

[0065] If the first detection result, the second detection result and the third detection result cannot be obtained at the same time within a preset number of consecutive frames, a danger instruction is generated; at the same time, detection is continued, and if the first detection result, the second detection result and the third detection result can be obtained at the same time, the danger instruction is updated to a safety instruction.

[0066] Furthermore, the control instruction generation module 16 in the limited space operation supervision system assisted by AI vision is also used for:

[0067] The harmful gas concentration in the target area is monitored in real time through the harmful gas sensor built into the wearable device. If the harmful gas concentration exceeds the preset concentration, a first alarm instruction is generated; the noise value in the target area is monitored in real time through the noise sensor. If the noise value exceeds the preset noise value, a second alarm instruction is generated; the control instruction is generated according to the first alarm instruction, the second alarm instruction and the risk assessment result.

[0068] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1 The limited space operation supervision method and specific examples of the auxiliary AI vision in the first embodiment are also applicable to the limited space operation supervision system of the auxiliary AI vision in this embodiment. Through the above detailed description of the limited space operation supervision method of the auxiliary AI vision, those skilled in the art can clearly know the limited space operation supervision system of the auxiliary AI vision in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0069] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0070] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A limited space operation supervision method assisted by AI vision, characterized in that: The AI ​​vision-assisted limited space operation supervision method is applied to the AI ​​vision-assisted limited space operation supervision system, the AI ​​vision-assisted limited space operation supervision system is communicatively connected with the safety specification detection system, the safety specification detection system includes an image acquisition device, a video server, an industrial computer, and an alarm, and the AI ​​vision-assisted limited space operation supervision method includes: Using the image acquisition device to collect high-definition operation image data of the target area in real time; The high-definition operation image data is compressed and transcoded by the video server to generate packaged data; The packaged data is input into the industrial computer, wherein the industrial computer includes a cutting unit and a detection unit, and the cutting unit performs ROI cutting on the packaged data according to the working area to generate working area image data; The operation detection model deployed in the detection unit detects the operation area image data to obtain operation detection results, including a first detection result, a second detection result, and a third detection result; Combining the first detection result, the second detection result, and the third detection result, performing a risk assessment based on a preset rule to obtain a risk assessment result; A control instruction is generated according to the risk assessment result, and the alarm is controlled according to the control instruction.

2. The AI ​​vision-assisted limited space operation supervision method according to claim 1, characterized in that: The image acquisition device also includes a high-definition camera, a WIFI transmitter, and a WIFI receiver. The high-definition camera is connected to the WIFI transmitter via PoE; the WIFI receiver is connected to the video server via a first preset interface; the video server is connected to the industrial computer via a second preset interface; and the industrial computer is connected to the alarm via a third preset interface.

3. The AI ​​vision-assisted limited space operation supervision method as claimed in claim 2, characterized in that: Also includes: The high-definition operating image data in the target area is collected along a preset collecting direction by a high-definition camera deployed at the top of the target area, and the collected high-definition operating image data is transmitted to the WIFI transmitter; The WIFI transmitter transmits the collected high-definition operation image data to the WIFI receiver by wireless transmission; The WIFI receiver transmits the high-definition operation image data to the video server.

4. The AI ​​vision-assisted limited space operation supervision method according to claim 1, characterized in that: The cropping unit performs ROI cropping on the packaged data according to the operating area to generate operating area image data, and further includes: Converting the packaged data into HSV image data; Creating a mask based on a preset color detection interval according to the HSV image data; Filling the mask with a closing operation, and then performing an opening operation on the filled data to obtain standard image data; Calculating the contour area of ​​the standard image data, and taking the largest contour area as the operation area; The packaged data is cropped according to the operation area to obtain the operation area image data.

5. The AI ​​vision-assisted limited space operation supervision method according to claim 1, characterized in that: The operation area image data is detected by using the operation detection model deployed in the detection unit, and further includes: Using LabelImg to label the image data of the working area frame by frame; Divide the labeled image data of the working area into a training set and a verification set according to a preset ratio, wherein the training set is 1600 frames and the verification set is 400 frames; Training the job detection model according to the training set to obtain a detection weight; Performing a performance evaluation on the detection weights according to the validation set, and selecting the optimal detection weights according to the evaluation results; The obtained detection weights are evaluated using a validation dataset of a limited workspace image dataset to quantify the weight detection accuracy and screen out the optimal detection weights; The operation area image data is detected according to the TensorRT model, the optimal detection weight and the operation detection model.

6. The AI ​​vision-assisted limited space operation supervision method according to claim 1, characterized in that: Combining the first detection result, the second detection result, and the third detection result to perform risk assessment based on preset rules to obtain a risk assessment result, further comprising: If the first detection result, the second detection result and the third detection result cannot be obtained at the same time within a preset number of consecutive frames, a danger instruction is generated; Meanwhile, detection is continuously performed, and if the first detection result, the second detection result and the third detection result can be obtained at the same time, the dangerous instruction is updated to a safe instruction.

7. The AI ​​vision-assisted limited space operation supervision method according to claim 1, characterized in that: Generating a control instruction according to the risk assessment result, and controlling the alarm according to the control instruction, further comprising: The harmful gas concentration in the target area is monitored in real time by using a harmful gas sensor built into the wearable device, and if the harmful gas concentration exceeds a preset concentration, a first alarm instruction is generated; The noise value in the target area is monitored in real time by a noise sensor, and if the noise value exceeds a preset noise value, a second alarm instruction is generated; The control instruction is generated according to the first alarm instruction, the second alarm instruction and the risk assessment result.

8. AI vision-assisted limited space operation supervision system, characterized in that: The steps for implementing the AI ​​vision-assisted confined space operation supervision method according to any one of claims 1 to 7, wherein the AI ​​vision-assisted confined space operation supervision system comprises: A data acquisition module, used to collect high-definition operation image data of the target area in real time using an image acquisition device; An image processing module, used for compressing and transcoding the high-definition operation image data through a video server to generate packaged data; An image cropping module, used for inputting the packaged data into an industrial computer, wherein the industrial computer comprises a cropping unit and a detection unit, wherein the cropping unit performs ROI cropping on the packaged data according to the operating area to generate operating area image data; An image detection module, configured to detect the image data of the operation area through an operation detection model deployed in the detection unit to obtain operation detection results, including a first detection result, a second detection result, and a third detection result; A risk assessment module, configured to perform risk assessment based on preset rules in combination with the first detection result, the second detection result, and the third detection result to obtain a risk assessment result; A control instruction generating module is used to generate a control instruction according to the risk assessment result, and control the alarm according to the control instruction.

Images