A method and system for underground safety detection in coal mines based on facial image recognition

Abstract

This invention discloses a method and system for underground safety detection in coal mines based on facial image recognition, belonging to the field of coal mine safety detection technology. The specific implementation steps include: identifying the miners' identities and safety equipment wearing status in each work area using video target recognition technology and adaptive facial recognition methods, obtaining miner numbers and a first hazard index; secondly, analyzing the hazard level of various miners' behaviors based on motion trajectory matching and behavior analysis technology, obtaining a second hazard index; then, assessing the hazard level of the miners' working environment based on environmental parameters, calculating a third hazard index; finally, combining the miners' hazard index and the hazard level of the work area, comprehensively assessing the miners' work safety situation, obtaining a comprehensive hazard index, and classifying the hazard level to issue timely warning information, effectively reducing the safety risks of underground coal mine operations, ensuring the safety of workers, and reducing the occurrence of safety accidents.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety detection technology, specifically to a method and system for underground coal mine safety detection based on facial image recognition. Background Technology

[0002] Coal mines, with their confined underground working spaces, are prone to accidents such as gas explosions, coal dust explosions, mine water hazards, and roof falls, posing serious threats to the lives of workers and mine production. With the increasing depth and scale of coal mining, these safety issues are becoming more frequent. Therefore, underground safety monitoring technology is particularly important. This technology, through real-time monitoring of underground environmental parameters such as gas concentration, carbon monoxide content, temperature, humidity, and wind speed, can promptly detect potential safety hazards and provide early warning information. It not only provides crucial technical support for mine safety management but also effectively improves emergency response capabilities and accident handling efficiency, significantly reducing the probability of safety accidents.

[0003] Currently, existing technologies for safety monitoring of miners are limited to detecting the miners' physical condition and dangerous behaviors. They ignore the interference of safety equipment (such as safety helmets and masks) on miner identification, and also fail to consider the impact of work fatigue and the degree of danger in the work area on miners' lives. As a result, the safety monitoring results cannot fully reflect the safety risks of mining operations, thus making it impossible to track the safety status of each miner in a timely manner, increasing the risk of operational errors and accidents.

[0004] To address this, a method and system for underground safety detection in coal mines based on facial image recognition is proposed. Summary of the Invention

[0005] This invention provides a method and system for underground safety monitoring in coal mines based on facial image recognition, used for comprehensive monitoring of the safety status of personnel underground. To address problems in existing technologies, this invention uses video target recognition technology and adaptive facial recognition to identify the identity of miners and their safety equipment wearing status in each work area, obtaining miner numbers and a first hazard index. Secondly, based on motion trajectory matching and behavior analysis technology, the hazard level of various miner behaviors is analyzed to obtain a second hazard index. Then, based on environmental parameters, the hazard level of the miner's working environment is assessed, calculating a third hazard index. Finally, combining the miner's hazard index and the hazard level of the work area, a comprehensive assessment of the miner's work safety is conducted, resulting in a comprehensive hazard index and hazard level classification. This allows for timely issuance of early warning information, effectively reducing safety risks in underground coal mine operations, ensuring the safety of workers, and reducing the occurrence of accidents.

[0006] A method for underground safety detection in coal mines based on facial image recognition, comprising the following specific implementation steps:

[0007] Based on the monitoring cameras, inspection robots, and sensors in each working area of ​​the target coal mine, real-time monitoring video data, inspection video data, and environmental parameters are collected; based on the facial recognition equipment at the entrance of the target coal mine, the first facial image and identity information data of the miners underground are obtained;

[0008] Data cleaning is performed on the monitoring video data and the inspection video data of each of the work areas to obtain first video data and second video data;

[0009] Video target recognition technology is used to acquire video data of all personnel in the second video data; the wearing status of safety equipment in the personnel video data is classified using a target recognition method; and the miner classification video data is identified using an adaptive face recognition method to obtain the miner number and the first danger index.

[0010] The identities of the people in the first video data are matched using a motion trajectory matching method; behavioral analysis is then performed on the matched first and second video data to obtain a second danger index.

[0011] Based on the environmental parameters, environmental assessments are conducted on different work areas to obtain a third hazard index;

[0012] By combining the first hazard index, the second hazard index, and the third hazard index, the work safety situation of the miners is comprehensively assessed to obtain a comprehensive hazard index; based on the comprehensive hazard index of different miners, the miners are classified into hazard levels.

[0013] Preferably, the environmental parameters include methane concentration index, carbon monoxide concentration, dust index, temperature index, humidity index, noise index, and vibration index; the first facial image represents the facial image of the target coal mine miner before work; the identity information data includes the miner's ID number and a set of facial images.

[0014] Preferably, the specific implementation process of the adaptive face recognition method includes:

[0015] The process involves acquiring the miner classification video data, including first-category videos, second-category videos, third-category videos, and fourth-category videos; combining the multi-regional face recognition unit and the identity information data to perform face recognition on the first-category videos, second-category videos, and third-category videos to obtain miner IDs; obtaining the miner IDs of individuals in the fourth-category videos based on the first facial image and eyebrow / eye parameter estimation unit; setting different risk weights for different categories of videos; and obtaining the first risk index for different miners based on their safety equipment wearing status.

[0016] Preferably, the process of determining the miner's ID number using the multi-region facial recognition unit includes:

[0017] During the training phase, based on the identity information data, a facial dataset of miners is obtained. Through a data augmentation layer, each image in the facial dataset is divided into m×n equal-sized image blocks, and k of these blocks are subjected to g types of occlusion to obtain an augmented facial dataset. This augmented dataset is then input into a feature extraction layer to obtain facial features. Through a feature enhancement layer, the texture feature weights of these facial features are obtained, and after feature reconstruction, enhanced facial features are obtained. These enhanced facial features are then input into a classifier to obtain facial recognition results. Finally, the facial recognition results are supervised using the angle boundary loss function and the cross-entropy loss function to obtain the trained multi-region face recognition unit.

[0018] During the inference phase, the first category video, the second category video, and the third category video are input into the trained multi-region face recognition unit to determine the miner's identity; the miner's number is obtained by combining the identity information data.

[0019] Preferably, the process of obtaining the miner's number based on the first facial image and the eyebrow and eye parameter estimation unit includes:

[0020] Using a facial landmark detection algorithm, the frontal eyebrow and eye landmarks of the first facial image are obtained. Based on the coordinates of the frontal eyebrow and eye landmarks, first eyebrow and eye parameters are obtained, including inter-eye distance, eye aspect ratio, eye opening angle, eyebrow aspect ratio, eyebrow and eye distance, and eyebrow curvature angle. Using the facial landmark detection algorithm, multi-view eyebrow and eye landmarks of the fourth category of videos are obtained. Using a head pose estimation algorithm, the Euler angles of each video frame in the fourth category of videos are obtained. Combining the multi-view eyebrow and eye landmarks and the Euler angles of the corresponding video frames, second eyebrow and eye parameters are obtained. Based on cosine similarity and mean absolute distance, the matching degree between the second eyebrow and eye parameters and the first eyebrow and eye parameters is obtained. If the matching degree is greater than a predetermined matching degree, the miner number is determined by the first eyebrow and eye parameter corresponding to the highest matching degree. If the matching degree is less than or equal to the predetermined matching degree, it indicates that the person corresponding to the second eyebrow and eye parameter is not a miner.

[0021] Preferably, the process of performing behavioral analysis on the matched first video data and second video data to obtain the second danger index includes:

[0022] The second video data is input to the eye-closing behavior detection unit to obtain the eye-closing duration and frequency corresponding to the miner's ID; the second video data is input to the head posture estimation unit to obtain the miner's Euler angles; the change amplitude of the miner's head is obtained by the change of Euler angles in the video; if the change amplitude of the head is greater than a predetermined amplitude threshold, it indicates that the miner has head-shaking behavior; the head-shaking frequency corresponding to the miner's ID is obtained based on the head-shaking behavior of the miner in the second video data; the first video data is input to the body sway detection unit to obtain the body swaying frequency and body swaying amplitude through the center of gravity shift algorithm; according to the dangerous action recognition unit, the dangerous actions of the miner are identified through a spatiotemporal convolutional neural network to obtain the number of occurrences and the danger level of the dangerous actions; the second danger index corresponding to the miner's ID is obtained by combining the eye-closing danger weight, head-shaking danger weight, body-shaking danger weight, and dangerous action weight.

[0023] Preferably, the process of conducting environmental assessments for different work areas includes:

[0024] Each environmental parameter of each work area is compared with its corresponding predetermined safety threshold to obtain the degree and number of times the environmental parameter exceeds the corresponding predetermined safety threshold. If the degree of exceedance is greater than a predetermined boundary threshold or the number of exceedances is greater than a predetermined number threshold, the corresponding environmental parameter is at a first environmental hazard level. If the degree of exceedance is less than or equal to the predetermined boundary threshold or the number of exceedances is less than or equal to the predetermined number threshold, the corresponding environmental parameter is at a second environmental hazard level. Based on the environmental parameters and corresponding environmental hazard levels of all work areas, a third hazard index is obtained for each work area.

[0025] Preferably, the specific implementation process for comprehensively assessing the miners' work safety includes:

[0026] Obtain the work hazard weights for all work areas, and combine them with the third hazard index corresponding to the work area to obtain the work hazard index; based on the work area where each miner is located, obtain the work hazard index corresponding to the miner's number; combine the first hazard index, the second hazard index, and the work hazard index corresponding to the miner's number to obtain the comprehensive hazard index corresponding to all miner numbers.

[0027] Preferably, the process of classifying miners into risk levels based on their comprehensive risk index includes:

[0028] The miner's ID, the first hazard index, the second hazard index, and the third hazard index are sent to the management personnel. If the overall hazard index is greater than or equal to a predetermined first hazard threshold, the miner is in the first hazard level, and a red alert is issued. If the overall hazard index is less than the predetermined first hazard threshold but greater than a predetermined second hazard threshold, the miner is in the second hazard level, and the corresponding miner's ID and work area are sent to the management personnel, and a yellow alert is issued. If the overall hazard index is less than or equal to the predetermined second hazard threshold, the miner is in the third hazard level, and no alert is issued.

[0029] A coal mine underground safety detection system based on facial image recognition includes:

[0030] The data acquisition module collects real-time monitoring video data, inspection video data, and environmental parameters based on the monitoring cameras, inspection robots, and sensors in each working area of ​​the target coal mine; and obtains the first facial image and identity information data of the miners underground based on the facial recognition equipment at the entrance of the target coal mine.

[0031] The data cleaning module cleans the monitoring video data and the inspection video data of each of the work areas to obtain first video data and second video data.

[0032] The miner identification module acquires video data of all personnel in the second video data through video target recognition technology; classifies the wearing status of safety equipment in the personnel video data through target recognition method; and identifies the miner classification video data through adaptive face recognition method to obtain the miner number and first danger index.

[0033] The miner behavior analysis module uses a motion trajectory matching method to match the identities of the people in the first video data; it then performs behavior analysis on the matched first and second video data to obtain a second danger index.

[0034] The environmental assessment module performs an environmental assessment of the work area based on the environmental parameters to obtain a third hazard index;

[0035] The miner safety comprehensive assessment module combines the first hazard index, the second hazard index, and the third hazard index to comprehensively assess the miner's work safety situation and obtain a comprehensive hazard index; based on the comprehensive hazard index of different miners, the miners are classified into hazard levels.

[0036] The monitoring and early warning module sends the miner ID numbers and the hazard levels of all miners to the management personnel in real time; if a miner's hazard level reaches the first or second hazard level, an early warning is issued to the management personnel.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] 1. The adaptive face recognition method proposed in this invention significantly improves recognition accuracy in the complex environment of underground coal mines by combining a multi-region face recognition unit and an eyebrow and eye parameter estimation unit. When miners are wearing safety helmets or masks, the multi-region face recognition unit introduces face occlusion interference into the dataset through data augmentation strategies, thereby enhancing the ability to extract facial features. This effectively addresses the partial occlusion of faces by underground miners' safety equipment, improving the accuracy of face recognition. Furthermore, the simultaneous wearing of safety helmets and masks makes it difficult to extract sufficient facial features. The eyebrow and eye parameter estimation unit utilizes the matching of frontal and multi-view eyebrow and eye key points to identify miners even when only eyebrow and eye features are captured, further improving the accuracy of face recognition and enabling timely tracking of miners' safety status.

[0039] 2. The behavior analysis method proposed in this invention utilizes multiple techniques, including closed-eye behavior detection, head posture estimation, body sway detection, and dangerous action identification, to comprehensively monitor and analyze miners' behavior during underground operations. This method can not only capture miners' fatigue status in real time but also identify potential dangerous actions, thereby accurately assessing the behavioral risks of miners. This helps to promptly detect and correct unsafe behaviors and prevent accidents.

[0040] 3. This invention comprehensively assesses the safety of miners' work by combining facial recognition results, behavioral analysis results, and environmental assessment results to form a comprehensive evaluation of miners' work safety. Simultaneously, by incorporating the hazard weights of the work area, a comprehensive hazard index for each miner is calculated. This not only provides managers with an intuitive basis for safety management but also makes safety management decisions more scientific and rational. Managers can take targeted management measures based on the different hazard levels of different miners, such as strengthening training, adjusting job positions, or conducting focused monitoring, thereby effectively improving the efficiency of safety management in coal mines. Attached Figure Description

[0041] Figure 1 A flowchart illustrating a coal mine underground safety detection method based on face image recognition, provided as an embodiment of the present invention;

[0042] Figure 2 This is a schematic diagram illustrating different image occlusion methods in the multi-region face recognition unit provided in an embodiment of the present invention;

[0043] Figure 3 A flowchart for miner behavior analysis provided in an embodiment of the present invention;

[0044] Figure 4This is a flowchart of a coal mine underground safety detection system based on face image recognition, provided as an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] The underground working environment in coal mines is complex and changeable, often accompanied by major safety hazards such as gas explosions, collapses, and water inrushes, posing a significant threat to the lives of underground workers. Furthermore, due to the high intensity and long working hours, miners are prone to fatigue, which reduces their reaction time and judgment, increasing the risk of accidents. Therefore, introducing underground personnel safety monitoring methods can monitor the location, behavior, and key parameters of the surrounding environment of underground workers in real time. When abnormalities are detected, timely warning signals can be issued, helping managers to quickly take emergency measures, reducing the risk of accidents, and effectively improving the safety of underground workers.

[0047] This invention proposes a method and system for underground safety detection in coal mines based on facial image recognition to detect the safety of personnel underground and ensure the safety of miners. To illustrate the effectiveness of the method and system, detailed descriptions will be provided in conjunction with the accompanying drawings and the following two embodiments.

[0048] Example 1

[0049] This application discloses a method for underground safety detection in coal mines based on facial image recognition, to detect the safety status of personnel underground in Coal Mine A. (See attached document.) Figure 1 The specific steps of the method proposed in this invention include: S1. Collecting video data and environmental data from each work area to obtain the miner's first facial image and identity information data; S2. Cleaning the video data from each work area; S3. Using an adaptive face recognition method, identifying video data of miners wearing safety equipment under different conditions to obtain the miner's number and a first danger index; S4. Analyzing the miner's behavior in the video data to obtain a second danger index; S5. Conducting environmental assessments of different work areas to obtain a third danger index; S6. Comprehensively assessing the miner's work safety situation to obtain a comprehensive danger index and classifying the miner into danger levels.

[0050] Furthermore, the working area underground in Coal Mine A is divided into k working areas, including ordinary working areas and hazardous working areas; each working area is equipped with at least one monitoring camera, at least one inspection robot, and several sensors; based on the monitoring cameras, inspection robots, and sensors in each working area, monitoring video data, inspection video data, and environmental parameters are collected in real time; according to the facial recognition equipment at the entrance of the target coal mine, the first facial image and identity information data of the underground miner are obtained, corresponding to step S1 above; wherein, the environmental parameters include gas concentration index, carbon monoxide concentration, dust index, temperature index, humidity index, noise index, and vibration index; the first facial image represents the facial image of the underground miner of the target coal mine before work; the identity information data includes the miner's number and facial image set.

[0051] This application embodiment provides a comprehensive information foundation for subsequent safety inspections by collecting real-time monitoring videos and environmental parameters; it not only helps to promptly detect potential safety hazards and reduce the probability of accidents, but also enables the rapid identification of miners through facial recognition technology and real-time tracking of the safety status of each miner.

[0052] Further, the monitoring video data and the inspection video data of each of the work areas are cleaned to obtain first video data and second video data, corresponding to step S2 above; wherein, the data cleaning includes operations such as noise reduction, defogging, low light enhancement, glare removal and grayscale conversion.

[0053] Specifically, in the underground working environment of Coal Mine A, monitoring video data is often affected by various factors, such as coal dust, humidity, and low light, leading to a decline in image quality. Noise reduction can eliminate environmental noise, improve image clarity, and help to more accurately identify miners and safety equipment. Defogging technology can eliminate blurring caused by moisture or coal dust, making the monitoring image clearer. Since the working environment often has low light conditions, especially in areas far from light sources or equipment malfunctions, monitoring video images may be blurry and details may be lost. Therefore, low-light enhancement can improve visibility under low-light conditions, ensuring effective monitoring of personnel status even in dim environments. Furthermore, the reflection of light from miners' lamps can cause glare in the monitoring video, thus obscuring important safety information, such as the miners' safety equipment wearing status and potential dangerous behaviors. Therefore, glare reduction can reduce interference from light reflection, ensuring the accuracy of video information. Grayscale processing simplifies image data, reduces computational complexity, and thus accelerates the subsequent analysis and recognition process.

[0054] Further, video target recognition technology is used to acquire video data of all personnel in the second video data; the wearing status of safety equipment in the personnel video data is classified using a target recognition method to obtain the miner classification video data, including first category video, second category video, third category video, and fourth category video; an adaptive face recognition method is used to identify the miner classification video data to obtain the miner number and first danger index, corresponding to step S3 above; the specific implementation process of the adaptive face recognition method includes:

[0055] Combining the multi-regional facial recognition unit and the identity information data, facial recognition is performed on the first category of videos, the second category of videos, and the third category of videos to obtain miner IDs; based on the first facial image and the eyebrow and eye parameter estimation unit, the miner IDs of the personnel in the fourth category of videos are obtained; different risk weights are set for the first category of videos, the second category of videos, the third category of videos, and the fourth category of videos, and the first risk index of different miners is obtained based on the miners' safety equipment wearing status, with the specific calculation formula as follows:

[0056]

[0057] in, The first danger index represents the miner with miner number i; These represent the probabilities of miner i in the first, second, third, and fourth categories of videos, respectively. And each miner has only one α1, α2, α3, α4 represent the danger weights of the first, second, third, and fourth categories of videos, respectively, and 1>α1>α2≥α3>α4>0.

[0058] Specifically, the first category of videos consists of video data of personnel not wearing safety helmets and masks; the second category consists of video data of personnel wearing only masks; the third category consists of video data of personnel wearing only safety helmets; and the fourth category consists of video data of personnel wearing both safety helmets and masks. In the underground working environment of Coal Mine A, the wearing of safety equipment directly affects the safety of miners. Categorizing the wearing of safety equipment helps identify which workers are not wearing the necessary protective equipment as required, thus allowing for timely measures to reduce safety risks. Among these, miners in the first category of videos are at the highest risk level because they lack basic head and face protection and are vulnerable to direct injury from collapses, flying objects, or harmful gases. While miners in the second category of videos wear masks for respiratory protection, the lack of helmets means their heads are still exposed to the risk of physical injury, resulting in a relatively high level of danger. Miners in the third category of videos have head protection, but the absence of masks may expose them to harmful gases or dust, posing a health risk. Miners in the fourth category of videos have the highest safety level, possessing comprehensive protective measures and experiencing the lowest level of danger.

[0059] This application employs a method combining video target recognition and adaptive facial recognition technology to perform detailed analysis of mining operation videos, significantly improving mine safety management, effectively reducing the occurrence of safety accidents, and greatly protecting the lives of miners. This method can accurately identify all personnel in the video and automatically classify and assess their safety equipment wearing status, quickly identifying situations where safety equipment is not worn or is worn improperly. By setting different hazard weights for each category, it obtains the real-time hazard index for each miner, not only greatly improving safety supervision efficiency but also encouraging miners to consciously comply with safety regulations, reducing potential accidents caused by human negligence.

[0060] Furthermore, the process of determining the miner's ID number using the multi-region facial recognition unit includes:

[0061] During the training phase, based on the aforementioned identity information data, a facial dataset of the miners is obtained; see [link / reference]. Figure 2 The face dataset is augmented by dividing each image into m×n equal-sized image blocks, and g types of occlusion are applied to k of these blocks to obtain an augmented face dataset. This augmented face dataset is then input into a feature extraction layer to obtain facial features. A feature enhancement layer is used to obtain the texture feature weights of these facial features, and feature reconstruction is performed to obtain enhanced facial features. These enhanced facial features are then input into a classifier to obtain the face recognition result. The face recognition result is supervised using the angle boundary loss function and the cross-entropy loss function to obtain the trained multi-region face recognition unit. The specific calculation formula for the loss function is as follows:

[0062] L=β1·L arc +β2·L cro ;

[0063]

[0064] Where L represents the total loss function; L arc ,L cro These represent the angular boundary loss function and the cross-entropy loss function, respectively; N represents the number of samples; θ u ,θ v represents the angle between the facial enhancement features of the u-th and v-th samples and the texture feature weights of the corresponding real categories, respectively; s represents the magnification coefficient, used to enhance the discriminative power between different categories, obtained through fitting with a large amount of experimental data; m represents the angular boundary, used to enhance the similarity between the same categories, obtained through multiple experimental tests; C represents the number of categories, here referring to the total number of miners; y u ρ represents the true class of the u-th sample; u Let represent the probability that the u-th sample is predicted as the true class; log() represents the logarithmic function; cos() represents the cosine function; e () Represents a power function with the natural constant as its base;

[0065] During the inference phase, the first category video, the second category video, and the third category video are input into the trained multi-region face recognition unit to determine the miner's identity; the miner's number is obtained by combining the identity information data.

[0066] Specifically, in the underground of Coal Mine A, due to the complex and dangerous working environment, ensuring accurate identification of miners is crucial. By enhancing and extracting features from miners' facial data, the accuracy of facial recognition can be improved, ensuring accurate identification of miners even under different safety equipment wearing conditions. For example, wearing only a mask or only a safety helmet may cause occlusion and changes in facial features, making traditional facial recognition algorithms difficult to use effectively. Therefore, occlusion processing of training images during the training phase can increase the model's robustness to different wearing conditions, ensuring accurate identification of miners under various working conditions, thereby improving the effectiveness of safety management and risk assessment.

[0067] This application's embodiments significantly enhance the accuracy and robustness of miner identification through a multi-regional face recognition unit, providing strong technical support for reducing safety accidents and ensuring miners' lives. This unit utilizes data augmentation techniques to generate a rich facial augmentation dataset, combining feature enhancement and angular boundary loss functions to effectively improve the model's ability to recognize miners' faces in complex environments. This ensures accurate identification of miners even under adverse conditions such as occlusion and changes in lighting, buying valuable time for timely rescue and greatly reducing the risk of safety accidents.

[0068] Furthermore, the process of obtaining the miner's ID based on the first facial image and the eyebrow and eye parameter estimation unit includes:

[0069] Using a facial landmark detection algorithm, the frontal eyebrow and eye landmarks of the first facial image are obtained. Based on the coordinates of these frontal eyebrow and eye landmarks, first eyebrow and eye parameters are obtained, including inter-eye distance, eye aspect ratio, eye opening angle, eyebrow aspect ratio, eyebrow and eye distance, and eyebrow curvature angle. Using the same facial landmark detection algorithm, multi-view eyebrow and eye landmarks from the fourth category of videos are obtained. Using a head pose estimation algorithm, the Euler angles of each video frame in the fourth category of videos are obtained. Based on the Euler angles of the corresponding video frames, the corresponding rotation matrix is ​​obtained, and the multi-view eyebrow and eye landmarks are mapped to the frontal view to obtain the corresponding pseudo-frontal view keypoint coordinates. Based on the pseudo-frontal view keypoint coordinates of different individuals in the fourth category of videos, the corresponding eyebrow and eye parameters are obtained, and the mean of different eyebrow and eye parameters is calculated to obtain second eyebrow and eye parameters. Based on cosine similarity and mean absolute distance, the matching degree between the second eyebrow and eye parameters and the first eyebrow and eye parameters is obtained, with the specific calculation formula as follows:

[0070]

[0071] Among them, MD oi This represents the degree of matching between the second eyebrow and eye parameters of the o-th person in the fourth category of videos and the first eyebrow and eye parameters of the miner with miner number i; A o B represents the vector of the second eyebrow and eye parameters of the o-th person in the fourth category of videos; i The vector represents the first eyebrow and eye parameters of miner i; |||| represents the vector modulo operation; γ1, γ2 represent the adjustment parameters and 0 < γ2 < γ1 < 1; χ represents the number of eyebrow and eye parameters and χ = 6;

[0072] If the matching degree is greater than the predetermined matching degree, the miner's number is determined by the first eyebrow parameter corresponding to the highest matching degree; if the matching degree is less than or equal to the predetermined matching degree, it means that the person corresponding to the second eyebrow parameter is not a miner; the work area and video data of the non-miner are transmitted to the management personnel, and a blue warning is issued.

[0073] In underground operations at Coal Mine A, miners often wear safety helmets and masks, obscuring most of their facial features. By extracting key facial features of the miners' eyebrows and eyes from their frontal view before work begins using a facial landmark detection algorithm, and combining this with multi-view eyebrow and eye landmarks from different work areas, the obstruction caused by equipment can be effectively overcome. Matching eyebrow and eye parameters allows for accurate identification of miners, ensuring real-time monitoring of workers in high-risk environments.

[0074] This application's embodiments combine the frontal eyebrow and eye parameters of the first facial image with multi-view eyebrow and eye parameters from the fourth category of video, improving the flexibility and accuracy of miner identification. This method can capture the eyebrow and eye features of miners from different perspectives and, combined with head posture information, accurately analyze the eyebrow and eye parameters from multiple perspectives, thereby quickly determining the miner's identity. This not only improves identification accuracy but also enhances the system's adaptability to complex working environments, ensuring accurate identification even when the miner is at different angles or in different postures.

[0075] Furthermore, the identities of the individuals in the first video data are matched using a motion trajectory matching method; behavioral analysis is then performed on the matched first and second video data to obtain a second danger index, corresponding to... Figure 1 Step S4; see Figure 3 The specific implementation process of the behavior analysis includes:

[0076] The second video data is input to the eye-closing behavior detection unit to obtain the eye-closing duration and frequency corresponding to the miner's ID; the second video data is input to the head posture estimation unit to obtain the miner's Euler angles; the change amplitude of the miner's head is obtained through the change of Euler angles in the video; if the change amplitude of the head is greater than a predetermined amplitude threshold, it indicates that the miner has head-shaking behavior; the head-shaking frequency corresponding to the miner's ID is obtained based on the head-shaking behavior of the miner in the second video data; the first video data is input to the body sway detection unit to obtain the body swaying frequency and body swaying amplitude through the center of gravity shift algorithm; according to the dangerous action recognition unit, the dangerous actions of the miner are identified through a spatiotemporal convolutional neural network to obtain the number of occurrences and the danger level of the dangerous actions; combining the eye-closing danger weight, head-shaking danger weight, body-shaking danger weight, and dangerous action weight, the second danger index corresponding to the miner's ID is obtained, and the specific calculation formula is as follows:

[0077]

[0078]

[0079] in, This represents the second danger index for miner number i; w1, w2, w3, w4 represent the weighting coefficients for different actions and are 0. <w1≤w2≤w3<w4<1; These represent the danger indexes of the miner with miner number i when performing actions such as closing eyes, shaking head, shaking body, and dangerous behavior.

[0080] The duration of the miner with miner number i in the second video having his eyes closed; T represents the total duration of the second video; f represents the frequency of eye closure for the miner with miner number i; eye δ1 and δ2 represent the normal eye-closing frequency; δ1 and δ2 represent the corresponding weighting coefficients and 0 < δ2 ≤ δ1 < 1. ψ represents the average amplitude of head shaking of the miner with miner number i; head This indicates the normal range of head shaking. f represents the head shaking frequency of the miner with miner number i; head ε1 and ε2 represent the normal head shaking frequency; ε1 and ε2 represent the corresponding weighting coefficients and 0 < ε2 ≤ ε1 < 1. ψ represents the average amplitude of body swaying for miner i. body This indicates the normal range of body swaying. f represents the frequency of the body swaying of the miner with miner number b; body ∈1, ∈2 represent the normal body swaying frequency; ∈1, ∈2 represent the corresponding weight coefficients and 0 < ∈2 ≤ ∈1 < 1; λ represents the number of times the miner with miner number i performs the u-th dangerous action; u This represents the risk level weight of the u-th dangerous behavior, where 0 < λ. u <1; m represents the total number of dangerous behaviors by miners.

[0081] During underground operations at Coal Mine A, miners may experience fatigue and dizziness after working for extended periods. By using a closed-eye behavior detection unit, a head posture estimation unit, and a body sway detection unit, the miners' fatigue and dizziness can be assessed in real time, thereby preventing dangers caused by fatigue and loss of consciousness.

[0082] This application's embodiments, by comprehensively utilizing motion trajectory matching and the analysis of multiple behaviors, can fully monitor the safety status of miners during operations, thereby effectively reducing the occurrence of safety accidents. This method can not only monitor the duration and frequency of miners closing their eyes, as well as the swaying of their head and body, in real time, but also accurately identify dangerous actions through advanced spatiotemporal convolutional neural networks, achieving comprehensive and in-depth analysis of miners' behavior, thus ensuring the safe operation of miners in complex and ever-changing working environments.

[0083] Furthermore, based on the aforementioned environmental parameters, environmental assessments are conducted on different work areas to obtain a third hazard index, corresponding to... Figure 1 Step S5; the process of conducting environmental assessments for different work areas includes:

[0084] Each environmental parameter of each work area is compared with its corresponding predetermined safety threshold to obtain the degree and number of times the environmental parameter exceeds the corresponding predetermined safety threshold. If the degree of exceedance is greater than a predetermined boundary threshold or the number of exceedances is greater than a predetermined number threshold, the corresponding environmental parameter is at a first environmental hazard level. If the degree of exceedance is less than or equal to the predetermined boundary threshold or the number of exceedances is less than or equal to the predetermined number threshold, the corresponding environmental parameter is at a second environmental hazard level. Based on the environmental parameters and corresponding environmental hazard levels of all work areas, a third hazard index for each work area is obtained, calculated using the following formula:

[0085]

[0086] in, The third hazard index represents the work area numbered x; These represent the probabilities of the work area numbered x under different environmental hazard levels. And each of the aforementioned work areas has only one =1; This represents the weighting coefficient of the u-th environmental parameter at different environmental hazard levels, and This indicates the total number of environmental parameters, which include gas concentration index, carbon monoxide concentration, dust index, temperature index, humidity index, noise index, and vibration index.

[0087] This application embodiment comprehensively monitors and evaluates environmental parameters in each working area, enabling timely detection and quantification of potential environmental risks. This provides strong data support for mine safety management and effectively reduces safety accidents caused by environmental factors. The method compares environmental parameters with predetermined safety thresholds to determine the hazard level of the environmental parameters, comprehensively considering the degree and frequency of exceedances. This ensures the comprehensiveness and accuracy of the evaluation results, thereby reducing the occurrence of safety accidents and protecting miners' lives.

[0088] Furthermore, by combining the first hazard index, the second hazard index, and the third hazard index, a comprehensive assessment of the miner's work safety is obtained, resulting in a comprehensive hazard index, corresponding to... Figure 1 The S5 steps, specifically the implementation process, include:

[0089] Obtain the work hazard weights for all work areas, and combine them with the third hazard index corresponding to each work area to obtain the work hazard index; based on the work area where each miner is located, obtain the work hazard index corresponding to the miner's ID; combine the first hazard index, the second hazard index, and the work hazard index corresponding to the miner's ID to obtain the comprehensive hazard index corresponding to all miner IDs. The specific calculation formula is as follows:

[0090]

[0091]

[0092] Among them, DI i This represents the overall risk factor for miner number i. WD i ω1, ω2, and ω3 respectively represent the first danger index, the second danger index, and the work danger index of the miner with miner number i; ω1, ω2, and ω3 respectively represent the weighting coefficients and 0 < ω1 ≤ ω2 ≤ ω3 < 1; This represents the total working time of the miner with miner number i; This indicates the duration of work in the work area designated as x; This indicates the total number of work areas; The third hazard index represents the work area numbered x; ξ x This represents the hazard weight of the work area numbered x, where 0 < ξ. x ≤1.

[0093] To further illustrate the role of the comprehensive hazard assessment proposed in this invention, a comparative example is provided here of the comprehensive hazard indices of miners in Coal Mine A over a single day. Table 1 lists the comprehensive hazard indices for different miners; where the weighting coefficients ω1 = 0.3, ω2 = 0.3, and ω3 = 0.4. According to the data in Table 1, the comprehensive hazard index of different miners is influenced by the first hazard index, the second hazard index, and the job hazard index. For example, miner KG01 has a first hazard index, a second hazard index, and a work hazard index of 0.17, 0.54, and 0.66, respectively; while miner KG02 has a first hazard index, a second hazard index, and a work hazard index of 0.47, 0.78, and 0.61, respectively. Given that their work hazard indices are similar, miner KG02 has a higher overall hazard index because KG01's first and second hazard indices are significantly lower than KG02's. This indicates that the degree of danger in a miner's work area cannot be equated with the safety risks faced by the miner themselves. A comprehensive assessment of the miner's physical safety should be based on a combination of their personal protective measures and behavioral patterns.

[0094] Table 1. Comprehensive Risk Index of Different Miners in Coal Mine A This application's embodiments combine the first, second, and third hazard indices with the work hazard weights for different work areas to comprehensively and objectively assess the work safety situation of miners. This method not only covers the individual safety behaviors of miners and environmental factors but also fully considers the degree of hazard inherent in the work area itself, ensuring the comprehensiveness and accuracy of the assessment results, thereby effectively preventing safety accidents.

[0095] Furthermore, the process of classifying miners into risk levels based on their comprehensive risk index includes:

[0096] The miner's ID, the first hazard index, the second hazard index, and the third hazard index are sent to the management personnel. If the overall hazard index is greater than or equal to a predetermined first hazard threshold, the miner is in the first hazard level, and a red alert is issued. If the overall hazard index is less than the predetermined first hazard threshold but greater than a predetermined second hazard threshold, the miner is in the second hazard level, and the corresponding miner's ID and work area are sent to the management personnel, and a yellow alert is issued. If the overall hazard index is less than or equal to the predetermined second hazard threshold, the miner is in the third hazard level, and no alert is issued.

[0097] This application's embodiments classify miners into different hazard levels based on a comprehensive hazard index and implement corresponding early warning measures. This achieves accurate identification and timely response to miner safety risks, effectively reducing the risk of safety accidents. For miners in the first hazard level, a red alert is immediately issued, along with a detailed report of each hazard index, ensuring that management personnel can intervene quickly and take emergency measures to prevent potential safety accidents. For miners in the second hazard level, the system issues a yellow alert and indicates their work area, allowing management personnel to pay targeted attention and supervision, and promptly prevent safety hazards.

[0098] In underground operations at Coal Mine A, miners' safety may be threatened due to inadequate personal protective equipment, work fatigue, dangerous behavior, and environmental risks. Accurately identifying the hazard level of each miner ensures that management can take timely and appropriate measures. A red alert indicates that the miner is in an extremely high-risk state, at which point work can no longer continue, and immediate mandatory evacuation or rest is necessary to prevent major accidents. A yellow alert alerts management to potential risks, potentially requiring real-time monitoring of the corresponding miner to prevent accidents. Furthermore, the real-time transmission of miner numbers and corresponding hazard indices helps management analyze the causes of safety risks, facilitating subsequent measures such as safety training and updating management systems.

[0099] This application embodiment achieves comprehensive detection of the safety status of miners underground in coal mines through an adaptive face recognition method, behavior analysis method, and comprehensive evaluation mechanism. The specific implementation process mainly includes the following steps: (i) identifying miners based on the different wearing conditions of their safety equipment; (ii) comprehensively detecting and analyzing miners' behavior; and (iii) comprehensively evaluating the miners' work safety status. This invention proposes an adaptive face recognition method, behavior analysis, and comprehensive hazard assessment for the above three processes. The adaptive face recognition method enables miner identification even when safety equipment is obscured, improving the reliability of safety detection. The behavior analysis enhances the real-time detection capability of miners' fatigue state and dangerous actions. The comprehensive hazard assessment further improves the scientific and rational nature of safety management decisions, helping to effectively reduce the risk of accidents.

[0100] Example 2

[0101] In Example 1, the method of the present invention enables safety detection of personnel underground in Coal Mine A. This application will describe a coal mine underground safety detection system based on facial image recognition proposed in this invention, see reference [link to relevant documentation]. Figure 4 The underground safety monitoring system for coal mines includes:

[0102] The data acquisition module collects real-time monitoring video data, inspection video data, and environmental parameters based on the monitoring cameras, inspection robots, and sensors in each working area of ​​the target coal mine; and obtains the first facial image and identity information data of the miners underground based on the facial recognition equipment at the entrance of the target coal mine.

[0103] The data cleaning module cleans the monitoring video data and the inspection video data of each of the work areas to obtain first video data and second video data.

[0104] The miner identification module acquires video data of all personnel in the second video data through video target recognition technology; classifies the wearing status of safety equipment in the personnel video data through target recognition method; and identifies the miner classification video data through adaptive face recognition method to obtain the miner number and first danger index.

[0105] The miner behavior analysis module uses a motion trajectory matching method to match the identities of the people in the first video data; it then performs behavior analysis on the matched first and second video data to obtain a second danger index.

[0106] The environmental assessment module performs an environmental assessment of the work area based on the environmental parameters to obtain a third hazard index;

[0107] The miner safety comprehensive assessment module combines the first hazard index, the second hazard index, and the third hazard index to comprehensively assess the miner's work safety situation and obtain a comprehensive hazard index; based on the comprehensive hazard index of different miners, the miners are classified into hazard levels.

[0108] The monitoring and early warning module sends the miner ID numbers and the hazard levels of all miners to the management personnel in real time; if a miner's hazard level reaches the first or second hazard level, an early warning is issued to the management personnel.

[0109] Specifically, in the monitoring and early warning module, the management personnel's handling process depends on the miner's hazard level and whether they are wearing safety equipment. When a miner without safety equipment is identified, their miner number and hazard level are sent to the management personnel in real time, providing data for subsequent targeted safety training. If the miner's hazard level reaches the first or second level, the management personnel will receive a red or yellow warning, prompting them to take immediate action to reduce safety risks. For example, a red warning means that the miner is at extremely high risk, and the management personnel need to intervene quickly, such as stopping operations, mandatory rest, and emergency evacuation. A yellow warning requires the management personnel to monitor the corresponding miner in real time to prevent safety accidents from occurring.

[0110] This application's embodiments construct a comprehensive and efficient miner safety monitoring system, significantly improving the level of safety management in coal mines and effectively reducing the occurrence of safety accidents. The system can collect monitoring videos, inspection videos, and environmental parameters in real time, ensuring the comprehensiveness and timeliness of the data. Through an adaptive facial recognition method, it accurately identifies miners' identities and the wearing of safety equipment, tracking the safety status of different miners in real time. Simultaneously, it performs multi-dimensional analysis of miner behavior and environmental factors, comprehensively assessing the miners' work safety situation and achieving accurate classification of hazard levels. Furthermore, the real-time feedback mechanism of the monitoring and early warning module ensures that management personnel can quickly respond to potential safety risks and take necessary intervention measures, thereby effectively preventing safety accidents and protecting the lives of miners.

[0111] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for underground safety detection in coal mines based on facial image recognition, characterized in that, include: Based on the monitoring cameras, inspection robots and sensors in each working area of ​​the target coal mine, real-time monitoring video data, inspection video data and environmental parameters are collected; Based on the facial recognition equipment at the entrance of the target coal mine, the first facial image and identity information data of the miners underground are obtained; Data cleaning is performed on the monitoring video data and the inspection video data of each of the work areas to obtain first video data and second video data; Video target recognition technology is used to obtain video data of all people in the second video data; The wearing status of safety equipment in the personnel video data is classified using a target recognition method, including a first category of videos, a second category of videos, a third category of videos, and a fourth category of videos; An adaptive face recognition method was used to identify miners from video data to obtain their numbers and the highest risk index. The process of obtaining the miner's ID includes: Using a facial landmark detection algorithm, the frontal eyebrow and eye landmarks of the first facial image are obtained. Based on the coordinates of the frontal eyebrow and eye landmarks, first eyebrow and eye parameters are obtained, including inter-eye distance, eye aspect ratio, eye opening angle, eyebrow aspect ratio, inter-eye distance, and eyebrow curvature angle. Using the facial landmark detection algorithm, multi-view eyebrow and eye landmarks of the fourth category of videos are obtained. Using a head pose estimation algorithm, the Euler angles of each video frame in the fourth category of videos are obtained. Combining the multi-view eyebrow and eye landmarks and the Euler angles of the corresponding video frames, second eyebrow and eye parameters are obtained. Based on cosine similarity and mean absolute distance, the matching degree between the second eyebrow and eye parameters and the first eyebrow and eye parameters is obtained. If the matching degree is greater than a predetermined matching degree, the miner's number is determined by the first eyebrow and eye parameter corresponding to the highest matching degree. If the matching degree is less than or equal to the predetermined matching degree, it indicates that the person corresponding to the second eyebrow and eye parameter is not a miner. The identities of the people in the first video data are matched using a motion trajectory matching method; behavioral analysis is then performed on the matched first and second video data to obtain a second danger index. Based on the environmental parameters, environmental assessments are conducted on different work areas to obtain a third hazard index; By combining the first hazard index, the second hazard index, and the third hazard index, the work safety situation of the miners is comprehensively assessed to obtain a comprehensive hazard index; based on the comprehensive hazard index of different miners, the miners are classified into hazard levels.

2. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The first facial image represents the facial image of the target coal mine miner before work; the identity information data includes the miner's ID number and a set of facial images.

3. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The specific implementation process of the adaptive face recognition method includes: The process involves acquiring the miner classification video data, including first-category videos, second-category videos, third-category videos, and fourth-category videos; combining the multi-regional face recognition unit and the identity information data to perform face recognition on the first-category videos, second-category videos, and third-category videos to obtain miner IDs; obtaining the miner IDs of individuals in the fourth-category videos based on the first facial image and eyebrow / eye parameter estimation unit; setting different risk weights for different categories of videos; and obtaining the first risk index for different miners based on their safety equipment wearing status.

4. The method for underground safety detection in coal mines based on facial image recognition according to claim 3, characterized in that, The process of determining the miner's ID number using the multi-region facial recognition unit includes: During the training phase, based on the identity information data, a facial dataset of miners is obtained; through a data augmentation layer, each image in the facial dataset is divided into... Five image patches of equal size, and among them... Each image block is processed A face augmentation dataset is obtained by occlusion; the face augmentation dataset is input into the feature extraction layer to obtain face features; the texture feature weights of the face features are obtained through the feature enhancement layer, and the face enhancement features are obtained through feature reconstruction; the face enhancement features are input into the classifier to obtain the face recognition result; the face recognition result is supervised according to the angle boundary loss function and the cross-entropy loss function to obtain the trained multi-region face recognition unit; During the inference phase, the first category video, the second category video, and the third category video are input into the trained multi-region face recognition unit to determine the miner's identity; the miner's number is obtained by combining the identity information data.

5. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The process of performing behavioral analysis on the matched first and second video data to obtain the second danger index includes: The second video data is input to the eye-closing behavior detection unit to obtain the eye-closing duration and frequency corresponding to the miner's ID; the second video data is input to the head posture estimation unit to obtain the miner's Euler angles; the change amplitude of the miner's head is obtained by the change of Euler angles in the video; if the change amplitude of the head is greater than a predetermined amplitude threshold, it indicates that the miner has head-shaking behavior; the head-shaking frequency corresponding to the miner's ID is obtained based on the head-shaking behavior of the miner in the second video data; the first video data is input to the body sway detection unit to obtain the body swaying frequency and body swaying amplitude through the center of gravity shift algorithm; according to the dangerous action recognition unit, the dangerous actions of the miner are identified through a spatiotemporal convolutional neural network to obtain the number of occurrences and the danger level of the dangerous actions; the second danger index corresponding to the miner's ID is obtained by combining the eye-closing danger weight, head-shaking danger weight, body-shaking danger weight, and dangerous action weight.

6. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The process of conducting environmental assessments for the different work areas includes: Each environmental parameter of each work area is compared with its corresponding predetermined safety threshold to obtain the degree and number of times the environmental parameter exceeds the corresponding predetermined safety threshold. If the degree of exceedance is greater than a predetermined boundary threshold or the number of exceedances is greater than a predetermined number threshold, the corresponding environmental parameter is at a first environmental hazard level. If the degree of exceedance is less than or equal to the predetermined boundary threshold or the number of exceedances is less than or equal to the predetermined number threshold, the corresponding environmental parameter is at a second environmental hazard level. Based on the environmental parameters and corresponding environmental hazard levels of all work areas, a third hazard index is obtained for each work area.

7. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The specific process for comprehensively assessing the work safety of the miners includes: Obtain the work hazard weights for all work areas, and combine them with the third hazard index corresponding to the work area to obtain the work hazard index; based on the work area where each miner is located, obtain the work hazard index corresponding to the miner's number; combine the first hazard index, the second hazard index, and the work hazard index corresponding to the miner's number to obtain the comprehensive hazard index corresponding to all miner numbers.

8. The method for underground safety detection in coal mines based on facial image recognition according to claim 1, characterized in that, The process of classifying miners into risk levels based on their comprehensive risk index includes: The miner's ID, the first hazard index, the second hazard index, and the third hazard index are sent to the management personnel. If the overall hazard index is greater than or equal to a predetermined first hazard threshold, the miner is in the first hazard level, and a red alert is issued. If the overall hazard index is less than the predetermined first hazard threshold but greater than a predetermined second hazard threshold, the miner is in the second hazard level, and the corresponding miner's ID and work area are sent to the management personnel, and a yellow alert is issued. If the overall hazard index is less than or equal to the predetermined second hazard threshold, the miner is in the third hazard level, and no alert is issued.

9. A coal mine underground safety detection system based on facial image recognition, characterized in that, include: The data acquisition module collects real-time monitoring video data, inspection video data, and environmental parameters based on the monitoring cameras, inspection robots, and sensors in each working area of ​​the target coal mine; and obtains the first facial image and identity information data of the miners underground based on the facial recognition equipment at the entrance of the target coal mine. The data cleaning module cleans the monitoring video data and the inspection video data of each of the work areas to obtain first video data and second video data. The miner identification module uses video target recognition technology to obtain video data of all personnel in the second video data. The wearing status of safety equipment in the personnel video data is classified using a target recognition method, including a first category of videos, a second category of videos, a third category of videos, and a fourth category of videos; An adaptive face recognition method was used to identify miners from video data to obtain their numbers and the highest risk index. The process of obtaining the miner's ID includes: Using a facial landmark detection algorithm, the frontal eyebrow and eye landmarks of the first facial image are obtained. Based on the coordinates of the frontal eyebrow and eye landmarks, first eyebrow and eye parameters are obtained, including inter-eye distance, eye aspect ratio, eye opening angle, eyebrow aspect ratio, inter-eye distance, and eyebrow curvature angle. Using the facial landmark detection algorithm, multi-view eyebrow and eye landmarks of the fourth category of videos are obtained. Using a head pose estimation algorithm, the Euler angles of each video frame in the fourth category of videos are obtained. Combining the multi-view eyebrow and eye landmarks and the Euler angles of the corresponding video frames, second eyebrow and eye parameters are obtained. Based on cosine similarity and mean absolute distance, the matching degree between the second eyebrow and eye parameters and the first eyebrow and eye parameters is obtained. If the matching degree is greater than a predetermined matching degree, the miner's number is determined by the first eyebrow and eye parameter corresponding to the highest matching degree. If the matching degree is less than or equal to the predetermined matching degree, it indicates that the person corresponding to the second eyebrow and eye parameter is not a miner. The miner behavior analysis module uses a motion trajectory matching method to match the identities of the people in the first video data; it then performs behavior analysis on the matched first and second video data to obtain a second danger index. The environmental assessment module performs an environmental assessment of the work area based on the environmental parameters to obtain a third hazard index; The miner safety comprehensive assessment module combines the first hazard index, the second hazard index, and the third hazard index to comprehensively assess the miner's work safety situation and obtain a comprehensive hazard index; based on the comprehensive hazard index of different miners, the miners are classified into hazard levels. The monitoring and early warning module sends the miner ID numbers and the hazard levels of all miners to the management personnel in real time; if a miner's hazard level reaches the first or second hazard level, an early warning is issued to the management personnel.

Images