Method and device for detecting safety of high-altitude operation

Abstract

The invention relates to a high-altitude operation safety detection method and device. The method comprises the following steps: acquiring a to-be-detected image of a high-altitude operation area; inputting the to-be-detected image into a visual detection model for detection to obtain a personnel detection result; wherein the detection result is used for determining whether an operator exists in the to-be-detected image; and under the condition that the operator exists in the to-be-detected image, safety equipment of the operator in the to-be-detected image is detected through a first detection algorithm and / or a second detection algorithm, and a safety detection result is obtained. Therefore, on one hand, the to-be-detected images are preliminarily detected, the to-be-detected images without operators are excluded, safety detection does not need to be carried out on all the to-be-detected images, safety detection resources are saved, and the detection efficiency is improved; on the other hand, high-altitude operation safety detection is carried out through two detection algorithms, and the accuracy and reliability of safety detection results are further guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safe construction detection, in particular to a detection method and device for high-altitude operation safety. BACKGROUND

[0002] The five common injuries in construction are falling from a height, object impact, mechanical injury, electric shock and collapse. The number of deaths caused by the five injuries accounts for more than 90% of the total number of deaths in the construction industry. Among them, the falling from a height accident has the highest incidence and is extremely dangerous.

[0003] Falling from a height accident can bring devastating blows to families and individuals. At present, the best way to prevent falling from a height is to wear personal protective equipment. However, the traditional manual inspection method for safety hazards is inefficient and difficult to find and eliminate potential risks. Therefore, it is urgent to improve the algorithm of intelligent equipment for detecting personal protective equipment to improve the detection efficiency.

[0004] Therefore, the prior art still needs to be improved and developed. SUMMARY

[0005] Therefore, it is necessary to provide a detection method and device for high-altitude operation safety in view of the above technical problems.

[0006] In a first aspect, the present application provides a detection method for high-altitude operation safety, which comprises:

[0007] obtaining a to-be-detected image of a high-altitude operation area;

[0008] inputting the to-be-detected image into a visual detection model for detection to obtain a personnel detection result; wherein the detection result is used to determine whether there is an operation personnel in the to-be-detected image;

[0009] in the case that there is the operation personnel in the to-be-detected image, detecting safety equipment of the operation personnel in the to-be-detected image by a first detection algorithm and / or a second detection algorithm to obtain a safety detection result.

[0010] In one of the embodiments, the method further comprises:

[0011] in the case that any of the safety detection results indicates that the high-altitude operation of the operation personnel is unsafe, outputting a warning information; wherein the warning information is used to remind the user to check the high-altitude operation safety in time.

[0012] In one of the embodiments, the first detection algorithm is a target detection model; and the detection of the safety equipment of the operation personnel in the to-be-detected image by the first detection algorithm to obtain the safety detection result comprises:

[0013] inputting the image to be detected into the target detection model to obtain the safety detection result; wherein a structure of the target detection model comprises an input end, a backbone network, a neck network and a prediction end;

[0014] The input end is configured to obtain a sample data set, perform mosaic data enhancement on sample images in the sample data set, perform adaptive anchor frame calculation on the sample data set to determine anchor frame sizes, and perform adaptive image scaling on the sample images to a predetermined size.

[0015] The backbone network is configured to extract features of the sample images layer by layer.

[0016] The neck network is configured to perform feature enhancement and feature fusion according to a feature pyramid and a gated convolution.

[0017] The prediction end is configured to output the safety detection result corresponding to the sample images.

[0018] In one of the embodiments, the detecting the image to be detected by the second detection algorithm to obtain the safety detection result comprises:

[0019] According to the current use state information of the safety equipment, a first safety parameter corresponding to the safety equipment is determined.

[0020] According to the current use information of the load-bearing frame of the operating personnel, a second safety parameter corresponding to the load-bearing frame is determined.

[0021] According to the first safety parameter and / or the second safety parameter, the safety detection result is determined.

[0022] In one of the embodiments, the safety equipment comprises a safety helmet; the current use state information comprises at least one of shell indentation information, shell damage information and suspension system use information; the shell indentation information comprises an indentation degree and an indentation number of the safety helmet; the shell damage information comprises a rupture degree and a rupture number of the safety helmet; and the suspension system use information comprises a headband wear degree and a suspension belt wear degree.

[0023] In one of the embodiments, the safety equipment comprises a safety belt; the current use state information comprises at least one of waistband use information, shoulder belt use information and back belt use information; the waistband use information comprises a waistband rupture degree; the shoulder belt use information comprises a shoulder belt rupture degree; and the back belt use information comprises a back belt rupture degree.

[0024] In one of the embodiments, the safety device comprises a safety glove; the current use state information comprises at least one of glove wear information and hole information; the glove wear information comprises at least one of a glove wear area and a glove wear degree; and the hole information comprises at least one of a number of holes and a hole area.

[0025] In one of the embodiments, the second safety parameter corresponding to the load-bearing frame is determined according to at least one of the current use information of the load-bearing frame of the worker.

[0026] The first alternative parameter is determined according to at least one of a theoretical service life, a remaining service life, a number of rusts, a rust area, a deformation degree, a cracking degree, and a number of cracks of the load-bearing frame.

[0027] The second alternative parameter is determined according to the weight of the worker, the number of workers, and a maximum load-bearing value of the load-bearing frame.

[0028] The second safety parameter is determined according to the first alternative parameter and / or the second alternative parameter.

[0029] In one of the embodiments, the safety detection result is determined according to the first safety parameter and / or the second safety parameter, including one of the following:

[0030] Firstly, the safety detection result is determined according to the first safety parameter.

[0031] Secondly, the safety detection result is determined according to the second safety parameter.

[0032] Thirdly, the safety detection result is determined according to a sum, an average, a ratio, and / or a variance of the first safety parameter and the second safety parameter.

[0033] Fourthly, the safety detection result is determined according to a product of the first safety parameter and a first influence factor, and a product of the second safety parameter and a second influence factor.

[0034] In a second aspect, the present application further provides a safety detection device for aerial work, comprising:

[0035] An acquisition module is configured to acquire a to-be-detected image of an aerial work area.

[0036] A first detection module is configured to input the to-be-detected image into a visual detection model for detection to obtain a personnel detection result, wherein the detection result is used to determine whether there is a worker in the to-be-detected image.

[0037] The second detection module is configured to detect the safety equipment of the worker in the to-be-detected image by using the first detection algorithm and / or the second detection algorithm to obtain a safety detection result when the worker exists in the to-be-detected image.

[0038] In a third aspect, the present application provides an electronic device. The electronic device comprises a memory for storing a processor computer program and a processor; wherein the processor is configured to implement the high-altitude work safety detection method in any of the embodiments of the present application when executing the computer program.

[0039] In a fourth aspect, the present application provides a computer readable storage medium. The computer readable storage medium has a computer program stored thereon, and the computer program is executed by a processor to implement the high-altitude work safety detection method in any of the embodiments of the present application.

[0040] The present application provides a high-altitude work safety detection method, comprising the following steps: obtaining a to-be-detected image of a high-altitude work area; inputting the to-be-detected image into a visual detection model for detection to obtain a personnel detection result; wherein the detection result is used to determine whether a worker exists in the to-be-detected image; and when the worker exists in the to-be-detected image, detecting the to-be-detected image by using a first detection algorithm and / or a second detection algorithm to obtain a safety detection result. In this way, on the one hand, the present application performs preliminary detection on the to-be-detected image to exclude to-be-detected images without workers, so that safety detection is not required for all to-be-detected images, safety detection resources are saved, and detection efficiency is improved; on the other hand, two detection algorithms are used for high-altitude work safety detection, which further guarantees the accuracy and reliability of the safety detection result. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flowchart of a high-altitude work safety detection method according to an exemplary embodiment;

[0042] Figure 2 is a schematic diagram of a safety helmet according to an exemplary embodiment;

[0043] Figure 3 is a schematic diagram of a safety rope according to an exemplary embodiment;

[0044] Figure 4 is a schematic diagram of a safety belt according to an exemplary embodiment;

[0045] Figure 5 is a structure diagram of a YOLOv5s-g n Conv model according to an exemplary embodiment;

[0046] Figure 6 FIG. 1 is a flow diagram of a method for detecting aerial work safety according to an example embodiment;

[0047] Figure 7 FIG. 2 is a block diagram of a device for detecting aerial work safety according to an example embodiment;

[0048] Figure 8 FIG. 3 is an internal structure diagram of an electronic device according to an example embodiment. DETAILED DESCRIPTION

[0049] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0050] The terms "first", "second", "third" in the embodiments of the present application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second", "third" can explicitly or implicitly include at least one of the features. In the description of the present application, "at least one" is used to indicate one or more; "a plurality of" means at least two, for example two, three, etc., unless otherwise explicitly specified. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or equipment.

[0051] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily mutually exclusive of other embodiments. It is explicitly and implicitly understood that the embodiments described herein can be combined with other embodiments.

[0052] The method described in the embodiments of the present application can be applied to an electronic device and / or an application located in the electronic device; the electronic device can be any mobile terminal or fixed terminal. The terminal can refer to a device that provides voice and / or data connectivity for a user. Exemplarily, the terminal can be an Internet of Things terminal, such as a sensor device, a mobile phone or a "cellular" phone, and a computer with an Internet of Things terminal, for example, which can be a fixed, portable, pocket-sized, handheld, computer-built-in or vehicle-mounted device. Among them, the portable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The application can be but is not limited to various software programs, fast applications and / or applets.

[0053] In some embodiments, as shown in Figure 1 , Figure 1 a flow chart of a detection method for high-altitude operation safety. Specifically, the present embodiment provides a detection method for high-altitude operation safety, which comprises the following steps:

[0054] Step S100: Obtain the to-be-detected image of the high-altitude operation area.

[0055] In the embodiments of the present application, the high-altitude operation area refers to a high place where a falling accident is likely to occur within a falling height reference plane of 2 meters or more (including 2 meters).

[0056] In some embodiments, the high-altitude operation type can include but is not limited to at least one of edge operation, climbing operation, suspension operation, crossing operation and hole operation.

[0057] In one embodiment, the image acquisition device is set to take pictures of the workers and safety equipment in the high-altitude operation area, obtaining real-time to-be-detected images. Exemplarily, the to-be-detected image of the power safety construction detection area indicates the images obtained for detection and analysis in the process of power safety supervision and management. These images usually include photos or video frames of the power operation site, and the shooting content may contain power workers working, equipment used, environmental conditions, safety equipment (such as safety belts, safety ropes), etc. Through these images, it can be detected and evaluated whether there are safety violations on site to ensure the safety and standardization of the operation.

[0058] Step S200: input the to-be-detected image into a visual detection model for detection, and obtain a personnel detection result; wherein the detection result is used to determine whether there is a worker in the to-be-detected image.

[0059] In some embodiments, the visual detection model can be a model obtained by training and testing training data containing positive examples (existence of workers) and negative example scenes (non-existence of workers).

[0060] In some embodiments, the electronic device can identify and analyze the to-be-detected image through a human body detection algorithm in computer vision, for example, by calculating the gradient direction of each pixel point in the to-be-detected image through a Histogram of Oriented Gradient (HOG) feature, and counting the histogram of the gradient direction to form a feature vector.

[0061] Step S300: In the case that the to-be-detected image contains the work personnel, detecting the safety equipment of the work personnel in the to-be-detected image through the first detection algorithm and / or the second detection algorithm respectively to obtain a safety detection result.

[0062] In the embodiments of the present application, the first detection algorithm and the second detection algorithm can be any algorithm and / or model for detecting high-altitude work safety. For example, it can be a multi-modal detection model, a target detection model, a semantic segmentation model, etc. for detecting whether the high-altitude work behavior is in violation, whether the work personnel wears safety equipment, and whether the safety equipment is qualified.

[0063] In the embodiments of the present application, the safety equipment can include, but is not limited to, at least one of a safety helmet, a safety rope, a safety belt, safety gloves, protective clothing, and a protective net.

[0064] For example, as shown in Figure 2 , the safety helmet at least includes a shell and a suspension system, and the safety helmet can further include a chin strap.

[0065] For example, as shown in Figure 3 , the safety rope includes a rope material, an end fixing device, and a connector.

[0066] For example, as shown in Figure 4 , the safety belt includes a belt, a connector, and an emergency unlocking device.

[0067] In this way, on the one hand, the present application performs preliminary detection on the to-be-detected image to exclude the to-be-detected image without work personnel, so that safety detection resources are saved and detection efficiency is improved without the need for safety detection on all to-be-detected images. On the other hand, the two detection algorithms are used for high-altitude work safety detection, further ensuring the accuracy and reliability of the safety detection result.

[0068] In some embodiments, the method further includes:

[0069] In any of the safety detection results indicating that the high-altitude work of the work personnel is in an unsafe condition, outputting a warning information; wherein the warning information is used to remind the user to check the high-altitude work safety in time.

[0070] In the embodiments of the present application, the early warning information includes at least one of text information, voice information, image information, and color information.

[0071] In some embodiments, the output early warning information includes at least one of:

[0072] outputting the early warning information on a display screen of the electronic device;

[0073] outputting the early warning information based on a voice device of the electronic device;

[0074] sending the early warning information to a third-party platform, and the early warning information is used for display by the third-party platform.

[0075] For example, the electronic device generates a safety detection report and sends it to a user mailbox in the form of an email; the safety detection report includes the early warning information.

[0076] In some embodiments, the method further includes:

[0077] If the safety factor in the safety detection result is less than or equal to a first preset threshold, it is determined that the overhead work of the worker is in an unsafe state.

[0078] In some embodiments, the electronic device can set at least one safety factor interval, different safety factor intervals are bounded by different predetermined values, and different safety factor intervals correspond to different safety states.

[0079] For example, if the range of the safety factor is 0 to 100, it is divided into four safety factor intervals. The first predetermined value can be 25, the second predetermined value can be 50, and the third predetermined value can be 75. Then 0-25 is the first safety factor interval, indicating a high-risk state; 25-50 is the second safety factor interval, indicating a low-risk state; 50-75 is the third safety factor interval, indicating a sub-safe state; and 75-100 is the fourth safety factor interval, indicating a safe state.

[0080] In this way, the electronic device can obtain safety detection results through different detection algorithms, different detection algorithms may focus on different overhead work safety risk points, and through the combination of the two methods, comprehensive coverage of overhead work safety detection can be achieved, and security vulnerabilities can be reduced. In any safety detection result indicating that the overhead work of the worker is in an unsafe state, the early warning information is output, the reliability and accuracy of the overhead work safety detection are ensured, and errors caused by a single method are reduced.

[0081] In some embodiments, the first detection algorithm is a target detection model; and the safety detection result obtained by detecting the safety equipment of the worker in the to-be-detected image through the first detection algorithm includes:

[0082] inputting the image to be detected into the target detection model to obtain the safety detection result; wherein the structure of the target detection model comprises an input end, a backbone network, a neck network and a prediction end:

[0083] The input end is configured to obtain a sample data set, perform mosaic data enhancement on sample images in the sample data set, perform adaptive anchor box calculation on the sample data set to determine anchor box sizes, and perform adaptive image scaling on the sample images to a predetermined size.

[0084] The backbone network is configured to extract features of the sample images layer by layer.

[0085] The neck network is configured to perform feature enhancement and feature fusion according to a feature pyramid and a gated convolution.

[0086] The prediction end is configured to output the safety detection result corresponding to the sample images.

[0087] In the embodiments of the present application, the target detection model can be a YOLO model. The target detection model can simultaneously perform target detection and classification through a single forward propagation process. The detection principle of the target detection model is shown in the following steps:

[0088] 1) Grid division: the input image to be detected is divided into a fixed size grid, and each grid unit is responsible for the target in the image to be detected.

[0089] 2) Anchor box: a predefined anchor box is a boundary box used to predict the position and size of a target. Each grid unit predicts multiple anchor boxes to adapt to targets of different shapes and sizes.

[0090] 3) Feature extraction: the input image to be detected is subjected to feature extraction by a convolutional neural network (CNN). For example, the target detection model can use Darknet as its backbone network to extract high-level feature representations from the input image to be detected.

[0091] 4) Prediction box regression and class prediction: the anchor box corresponding to each grid unit is responsible for predicting the position and size of the bounding box of the target. At the same time, the grid unit is also responsible for predicting which class the target belongs to. This step produces 4 coordinates (x, y, width, height) and C+1 class prediction values, where C is the number of classes.

[0092] 5) Confidence prediction: each grid unit is also responsible for predicting the confidence of the target. Confidence is used to represent the degree of confidence of the target detection.

[0093] 6) Post-processing: screening and adjusting the multiple bounding boxes generated for each grid cell. Exemplarily, non-maximum suppression (NMS) can be used to remove the bounding boxes with too much overlap, and only keep the one with the highest confidence.

[0094] 7) Output result: the target detection result can include, but is not limited to, at least one of the class, position and confidence of the target. The target detection result can be used to identify and locate the detected target in the image to be detected.

[0095] Exemplarily, the target detection model can be YOLOv2, YOLOv3, YOLOv4 or YOLOv5, etc.

[0096] In one embodiment, the target detection model is YOLOv5s-g n Conv model, such as Figure 5 as shown in Figure 5 YOLOv5s-g n Conv model. The structure diagram of YOLOv5s-g n Conv is a lightweight version with the fewest number of neural network layers and the fastest inference speed; YOLOv5s-g n Conv model is suitable for real-time target detection in resource-constrained environments. YOLOv5s-g n Conv model is improved based on the YOLOv5s model and consists of an input end (Input), a backbone network (Backbone), a neck network (Neck) and a prediction end (Prediction, output end). Through the network visualization tool Netron, the YOLOv5s-g n Conv target detection algorithm can be clearly displayed.

[0097] In some embodiments, the electronic device captures image data of the workers in the elevated work area through an image acquisition device on site, and / or obtains image data of the workers in the elevated work area from the Internet through a crawler technology. The electronic device expands the image data as a sample data set through a data enhancement technique; the data enhancement technique indicates a technique for expanding the data set by transforming the training data; the data enhancement technique includes, but is not limited to, at least one of adjusting brightness, random flipping, random rotation, adding noise and Mosaic; the sample data set is used to train the target detection model.

[0098] In some embodiments, the electronic device can perform data labeling on each sample image in the sample data set to obtain a labeled sample data set; data labeling indicates adding labels and / or annotations to each sample image in the sample data set to facilitate the electronic device to understand and utilize the sample data set; data labeling can include, but is not limited to, the following categories:

[0099] ①Target box annotation: Use the annotation tool LabelImg to frame the location of the target in the image. The target box is usually represented by a rectangle, including the coordinates of the top-left corner and the bottom-right corner of the target.

[0100] ②Class annotation: Assign the corresponding target class to each target box. For example, the annotation classes of the sample data set can be: safety helmet, safety belt, and safety rope.

[0101] ③Key point annotation: According to the need of the high-altitude operation of the workers, the sample data set can be annotated with the operation information of the workers. For example, the state of the worker at present can be estimated by human pose estimation.

[0102] ④Multi-target annotation: Handle the case of multiple targets in the image, and ensure that each target is correctly annotated. Samples with insufficient or unclear pixel area are not marked to prevent overfitting of the neural network.

[0103] ⑤Occlusion annotation: Handle the case where the target is occluded in the image. This can be represented by annotating the occlusion area or adjusting the shape of the target box. In the case of occlusion, targets with an occlusion area greater than 80% and targets with an image edge area less than 20% are not marked.

[0104] In some embodiments, the electronic device can perform a random shuffle operation on the annotated sample data set to ensure that the order of the samples in the sample data set is random, avoiding the dependence of the model on the order of the samples. The sample data set can be divided into a training set, a validation set, and a test set as needed. According to the preset ratio (for example, the preset ratio can be 7:2:1), the number of samples (images) that each data set should contain is calculated.

[0105] In one embodiment, the YOLOv5s-g n The input end of the Conv model performs Mosaic data augmentation on the sample data set, randomly selects four pictures as the original pictures to be spliced, randomly scales and crops each picture to adapt to the input size of the model, and then randomly distributes them for splicing. In this way, the detection data set is enriched, especially the random scaling increases a lot of small targets, improving the robustness of the network. Moreover, the input end performs adaptive anchor box calculation, which can use the initial anchor boxes of the COCO (Common Objects in Context) data set. The network outputs the prediction box based on the initial anchor box, compares it with the real box, calculates the difference between the two, and then updates the network parameters in reverse. The input end adaptively scales the data set of different sizes to an image of 608*608*3, and then sends it to the detection network.

[0106] In the embodiments of the present application, the Mosaic data enhancement can simulate the uncertainty of the position of the high-altitude worker behavior in the image and the characteristics of the light change in the interactive area by means of geometric deformation, random splicing, light and color adjustment and the like on the image to be detected, and randomly adding edge blank, so as to reduce the overfitting risk and enhance the model robustness.

[0107] In one embodiment, the YOLOv5s-g n The backbone network of the Conv model is composed of Focus, CSP1_X and Spatial Pyramid Pooling (SPP) network. First, the input image of 608*608*3 is sliced by the Focus network to reduce the height and width of the image, and is integrated by the Concatenate (Concat) operation to increase the channel number of the input image. At this time, the image channel number is 64. Then, the integrated image is subjected to feature extraction by the Conv convolution module with a size of 3 and a step of 2, and the output image size is 152*152*128. Then, after 3 groups of BottleneckCSP1_3 and Conv convolution operations, a feature map with an image size of 19*19*1024 is obtained. Finally, the SSP module is used to perform 1*1, 5*5, 9*9 and 13*13 maximum pooling operations on the 19*19 feature map to extract features from multiple aspects, and the four groups of pooled feature maps are aggregated by Concat to improve the model precision.

[0108] In one embodiment, the YOLOv5s-g n The neck network of the Conv network structure adopts the structure of FPN (Feature Pyramid Network) and PAN (Path Aggregation Network) for multi-scale fusion of features. Gated Convolution (g n Conv) replaces the spatial convolution in FPN for feature fusion, and g n Conv replaces the 3*3 convolution on the top-down path. Since the YOLOv5s target detection algorithm uses a large number of convolution models, g n Conv is added after the fusion of features of different pyramid levels, and g n Conv introduces g

[0109] In this embodiment, gated convolution typically refers to introducing a gating mechanism into a convolutional neural network (CNN) to improve the network's modeling ability and learning of long-range dependencies. This structure is inspired by gating units (such as LSTM and GRU) in recurrent neural networks (RNNs), but the gating mechanism is applied to the convolutional layers. This allows the convolutional layers to better handle sequential data or other data with temporal / spatial relationships.

[0110] The core idea of ​​gated convolution is to introduce a gating unit that dynamically and selectively retains or discards certain parts of the input feature map. This helps the network maintain focus on important information during learning and mitigates the impact of irrelevant information on the network. Below is a common implementation of gated convolution:

[0111] ① Gating Unit: A gating unit typically consists of two parts: an "input gate" and a "forget gate." These two gates control the flow of information from the input feature map.

[0112] ② Gating Mechanism: The input gate controls the transmission of input features, determining which features are important. It accepts the raw output of the convolutional layer as input and generates a value between 0 and 1, used to weight the raw output. The forget gate determines which information to retain. It also accepts the raw output of the convolutional layer and generates a value between 0 and 1, used to weight the raw output. Lower values ​​indicate more information to be forgotten.

[0113] ③ Gated convolution operation: The input feature map is multiplied by the product of the input gate and the forget gate to dynamically control the weights of the input features. In this way, the gated convolutional layer can adaptively and selectively weight different parts of the feature map by learning the parameters of the input gate and the forget gate.

[0114] Among them, g n The calculation of Conv can be divided into three parts:

[0115] The projection layers Φ in the first and last steps in and Φ out :

[0116] FLOPs(Φ in )=2HWC 2 FLOPs(Φ out =HWC 2 ; where Φ in and

[0117] Φ out It is a linear projection operation that completes the exchange of channel dimension information; H is the height of the image; W is the width of the image.

[0118] Depth-wise convolution f:

[0119] Depth-wise convolution with kernel size K*K acts on the feature above, where Therefore, the amount of calculation in this part is:

[0120]

[0121] The amount of calculation of g in dimension matching in the Recursive Gating (RG) operation:

[0122]

[0123] Therefore, the total amount of calculation is as follows:

[0124]

[0125] In some embodiments, the target detection model can be YOLOv5s-HorNet, which is based on the YOLOv5s model and replaces the g n Conv replaces the spatial mixing layer in vision Transformer or CNN, where the basic block contains a spatial mixing layer and a feed-forward network (FFN). According to the model size and depth, the convolution f k Implementation, there are two model variant series, named HorNet-T / S / B / L 7*7 and HorNet-T / S / B / L GF respectively. At the same time, Swin Transformer and ConvNeXt are regarded as vision Transformer and CNN baseline, because the model is implemented based on convolution framework, while having high-order interaction like vision Transformer. In order to make a fair comparison with the baseline, the number of blocks of Swin Transformers-S / B / L is followed, but an additional block is inserted in the 2nd stage to make the overall complexity close, resulting in [2, 3, 18, 2] blocks in each stage of all model variants. The electronic device only needs to adjust the basic number of channels C to build models of different sizes, and set the channel number of 4 stages as [C, 2C, 4C, 8C] according to the convention. For HorNet-T / S / B / L, C=64, 96, 128, 192 respectively. By default, the interaction order of each stage (i.e., g nn) is set to 2, 3, 4, 5, so that the channels of the coarsest stage C0 are the same in different stages. In this way, a network structure with horizontal connections is introduced, which can obtain more semantic information and contextual relationships by connecting feature maps of different levels, thereby helping to improve spatial interaction modeling and better capture relationship information between different targets. Since there are various interactions between workers and the environment (such as buildings, equipment, and personnel) in the construction site, horizontal connections can better understand the relationship between them. For example, horizontal connectivity can be better utilized to handle the interaction between safety equipment and workers and determine whether there are specifications for wearing personal safety (protective) equipment.

[0126] In some embodiments, the target detection model can be YOLOv5s-HorBlock, which adds g n Conv after fusing features of different pyramid levels. For target detection, replace the 3*3 convolution after the top-down path with g n Conv at each level layer. For semantic segmentation, simply replace the 3*3 convolution after concatenating multi-order feature maps with g n Conv. With the Hourglass network, target detection can be performed through multi-scale feature maps, further enhancing the perception and modeling capabilities of the model. By introducing horizontal connections, HorBlock establishes stronger contextual relationships in the model, thereby improving the ability to capture semantic information and better capturing relationships between targets, making the model more robust in complex environments; it can help solve the scale variation problem in construction sites, such as detecting targets near and far. Compared to YOLOv5s-HorBlock, it can also better handle occlusion in building structures and improve the model's ability to understand complex scenes.

[0127] In this way, the target detection model uses high-order spatial interaction to improve the accuracy of multi-scale target detection and recognition, and according to the different scales of the neck, it adapts to different sensory fields in each feature map to fuse safety equipment information from three branch receptive fields, improving the accuracy of multi-scale prediction. Using gated convolution in the convolution layer can enhance the model's ability to model local features. The gating mechanism helps the network to better selectively focus on different parts of the input data, thereby improving the ability to locate and recognize targets.

[0128] In some embodiments, the detecting the image by the second detection algorithm to obtain a safety detection result comprises:

[0129] According to the current use state information of the safety equipment, a first safety parameter corresponding to the safety equipment is determined;

[0130] According to the current use information of the load-bearing frame of the worker, a second safety parameter corresponding to the load-bearing frame is determined;

[0131] According to the first safety parameter and / or the second safety parameter, the safety detection result is determined.

[0132] In some embodiments, the safety device includes a safety helmet; the current use state information includes at least one of shell indentation information, shell breakage information, and suspension system use information; the shell indentation information includes indentation degree and indentation number of the safety helmet; the shell breakage information includes breakage degree and breakage number of the safety helmet; and the suspension system use information includes headband wear degree and suspension band wear degree.

[0133] In some embodiments, the determination of the first safety parameter according to the current use state information of the safety device includes:

[0134] The first safety parameter is determined according to the indentation degree, indentation number, breakage degree, breakage number, headband wear degree, and suspension band wear degree of the safety helmet.

[0135] Exemplarily, a way of determining the first safety parameter is: wherein T1 indicates the first safety parameter; M 凹 indicates the indentation degree of the safety helmet; M' 凹 indicates the maximum reference indentation degree of the safety helmet; S 凹 indicates the indentation number of the safety helmet; S' 凹 indicates the maximum reference indentation number of the safety helmet; M 破 indicates the breakage degree of the safety helmet; M' 破 indicates the maximum reference breakage degree of the safety helmet; S 破 indicates the breakage number of the safety helmet; S' 破 indicates the maximum reference breakage number of the safety helmet; M 头 indicates the headband wear degree; M' 头 indicates the maximum reference wear degree of the headband; M 悬 indicates the suspension band wear degree; M' 悬 indicates the maximum reference wear degree of the suspension band; and α1 to α6 indicate the weighted weights corresponding to the indentation degree, indentation number, breakage degree, breakage number, headband wear degree, and suspension band wear degree of the safety helmet. The weighted weights can be determined according to importance degree, influence degree, and / or authority setting.

[0136] In some embodiments, the safety equipment comprises a safety belt; the current usage state information comprises at least one of waist belt usage information, shoulder belt usage information, and back belt usage information; the waist belt usage information comprises a waist belt breaking degree; the shoulder belt usage information comprises a shoulder belt breaking degree; and the back belt usage information comprises a back belt breaking degree.

[0137] In some embodiments, the determining the first safety parameter corresponding to the safety equipment according to the current usage state information of the safety equipment comprises:

[0138] determining the first safety parameter according to the waist belt breaking degree, the shoulder belt breaking degree, and the back belt breaking degree.

[0139] Exemplarily, a manner of determining the first safety parameter is: wherein T1 indicates the first safety parameter; M 腰 indicates the waist belt breaking degree; M' 腰 indicates a maximum reference waist belt breaking degree; M 肩 indicates the shoulder belt breaking degree; M' 肩 indicates a maximum reference shoulder belt breaking degree; M 背 indicates the back belt breaking degree; M' 背 indicates a maximum reference back belt breaking degree; and a1 to a4 indicate weighted weights corresponding to the waist belt breaking degree, the shoulder belt breaking degree, and the back belt breaking degree. The weighted weights can be determined according to importance, influence, and / or authority.

[0140] In some embodiments, the safety equipment comprises a safety glove; the current usage state information comprises at least one of glove wear information and hole information; the glove wear information comprises at least one of a glove wear area and a glove wear degree; and the hole information comprises at least one of at least one of a hole number and a hole area.

[0141] In some embodiments, the determining the first safety parameter corresponding to the safety equipment according to the current usage state information of the safety equipment comprises:

[0142] determining the first safety parameter according to the glove wear area, the glove wear degree, the hole number, and the hole area.

[0143] Exemplarily, a manner of determining the first safety parameter is: wherein T1 indicates the first safety parameter; A 磨 indicates the glove wear area; A' 磨 indicates a maximum reference glove wear area; M 磨 indicates the glove wear degree; M' 磨 indicates a maximum reference glove wear degree; S 破indicates the number of holes of the safety gloves; S 破 indicates the maximum reference number of holes of the safety gloves; A 破 indicates the area of holes of the safety gloves; A' 破 indicates the maximum reference area of holes of the safety gloves; a1 to a4 indicate the weighted weights corresponding to the glove wear area, the glove wear degree, the number of holes and the area of holes of the safety gloves; the weighted weights can be determined according to the importance, the influence degree and / or the authority.

[0144] In this way, the first safety parameter of the safety equipment is determined by the current use state information of the plurality of safety equipment, the limitation and one-sidedness of the aerial work safety detection are reduced by considering from multiple dimensions, and the safety detection result can be determined more comprehensively and accurately. Moreover, in addition to detecting whether the work personnel wears the safety equipment, the influence of the damage of the safety equipment on the aerial work safety of the work personnel is further considered, thereby providing effective protection for the life safety of the work personnel.

[0145] In some embodiments, the second safety parameter corresponding to the load-bearing frame of the work personnel is determined according to the current use information of the load-bearing frame, including:

[0146] determining a first alternative parameter according to at least one of the theoretical service life, the remaining service life, the number of rusts, the area of rusts, the deformation degree, the cracking degree and the number of cracks of the load-bearing frame;

[0147] determining a second alternative parameter according to the weight, the number of the work personnel and the maximum load-bearing value of the load-bearing frame;

[0148] determining the second safety parameter according to the first alternative parameter and / or the second alternative parameter.

[0149] Exemplarily, a way of determining the first alternative parameter is: wherein, T 1备 indicates the first alternative parameter; M 变 indicates the deformation degree of the load-bearing frame; M' 变 indicates the maximum reference deformation degree of the load-bearing frame; S 锈 indicates the number of rusts of the load-bearing frame; S' 锈 indicates the maximum reference number of rusts of the safety helmet; M 裂 indicates the cracking degree of the load-bearing frame; M' 裂 indicates the maximum reference cracking degree of the load-bearing frame; A 锈 indicates the area of rusts of the load-bearing frame; A' 锈 indicates the maximum reference area of rusts of the load-bearing frame; S 裂 indicates the number of cracks of the load-bearing frame; S' 裂 indicates the maximum reference number of cracks of the load-bearing frame; L 理an indicator of a theoretical service life value of the load-bearing frame; L 剩 an indicator of a remaining service life value of the load-bearing frame; a1 to a6 are weighted weights corresponding to the deformation degree, the rust number, the cracking degree, the rust area, the cracking number, the theoretical service life and the remaining service life of the load-bearing frame; the weighted weights can be determined according to the importance degree, the influence degree and / or the authority setting.

[0150] In some embodiments, the determining the second security parameter according to the first alternative parameter and / or the second alternative parameter comprises one of the following:

[0151] The first kind: determining the second security parameter according to the minimum value of the first alternative parameter and the second alternative parameter;

[0152] The second kind: determining the second security parameter according to the first alternative parameter;

[0153] The third kind: determining the second security parameter according to the second alternative parameter;

[0154] The fourth kind: determining the second security parameter according to the product of the first alternative parameter and the third influence factor, and the product of the second alternative parameter and the fourth influence factor.

[0155] In this way, by understanding the service life of the load-bearing frame, the potential failure risk can be predicted, thereby avoiding high-altitude operation on the load-bearing frame close to or exceeding the designed service life, and reducing the possibility of structure collapse or failure; rust can seriously affect the load-bearing capacity and structural stability of the load-bearing frame, and by evaluating the rust degree, the rust problem can be found and handled in time to prevent further expansion of the rust leading to structural damage; by understanding the maximum load-bearing weight of the load-bearing frame, it can be ensured that the load-bearing capacity will not be exceeded during operation, thereby avoiding structural overload and potential safety risks.

[0156] In some embodiments, the determining the security detection result according to the first security parameter and / or the second security parameter comprises one of the following:

[0157] The first kind: determining the security detection result according to the first security parameter;

[0158] The second kind: determining the security detection result according to the second security parameter;

[0159] The third kind: determining the security detection result according to the sum, the mean, the ratio and / or the variance of the first security parameter and the second security parameter;

[0160] The fourth kind: determining the security detection result according to the product of the first security parameter and the first influence factor, and the product of the second security parameter and the second influence factor.

[0161] Therefore, the safety detection result can be reasonably and comprehensively evaluated by the safety parameters determined in different dimensions, and the accuracy and reliability of the safety detection of the aerial work are improved.

[0162] In the embodiments of the present application, the following specific examples are provided in combination with any of the above embodiments:

[0163] Specific example 1: Figure 6 An example of a safety detection method for aerial work is shown. As shown in Figure 6 The safety detection method for aerial work is executed by an electronic device, and the safety detection method for aerial work includes:

[0164] S601, obtaining a to-be-detected image of an aerial work area.

[0165] S602, inputting the to-be-detected image into a visual detection model for detection to determine whether there is a worker in the to-be-detected image.

[0166] In an optional embodiment, if it is determined that there is a worker in the to-be-detected image, S603 and S604 are entered; if it is determined that there is no worker in the to-be-detected image, S608 is entered.

[0167] S603, detecting the safety equipment of the worker in the to-be-detected image by a first detection algorithm to obtain a first safety detection result.

[0168] In an optional embodiment, the safety equipment of the worker in the to-be-detected image is detected by a target detection model to obtain the first safety detection result.

[0169] S604, detecting the safety equipment of the worker in the to-be-detected image by a second detection algorithm to obtain a second safety detection result.

[0170] In an optional embodiment, a first safety parameter corresponding to the safety equipment is determined according to the current use state information of the safety equipment; a second safety parameter corresponding to the load-bearing frame of the worker is determined according to the current use information of the load-bearing frame; and the second safety detection result is determined according to the first safety parameter and / or the second safety parameter.

[0171] S605, determining whether the aerial work of the worker is safe according to the first safety detection result and the second safety detection result.

[0172] In an optional embodiment, if the first safety detection result and the second safety detection result both indicate that the aerial work of the worker is in a safe condition, S606 is entered; if any safety detection result indicates that the aerial work of the worker is in an unsafe condition, S607 is entered.

[0173] S606, determine the high-altitude operation safety of the worker.

[0174] In an optional embodiment, the safety detection result is output on a display interface of the electronic device.

[0175] S607, determine that the high-altitude operation of the worker is unsafe.

[0176] In an optional embodiment, the safety detection result and the warning information are output on a display interface of the electronic device; wherein the warning information is used to remind the user to check the high-altitude operation safety in time.

[0177] S608, end.

[0178] The above high-altitude operation safety detection method and device, on the one hand, through preliminary detection on the to-be-detected image, the to-be-detected image without the worker is excluded, and safety detection is not needed for all to-be-detected images, safety detection resources are saved, and detection efficiency is improved; on the other hand, through two detection algorithms for high-altitude operation safety detection, the accuracy and reliability of the safety detection result are further ensured.

[0179] It should be understood that although each step in the flowchart involved in each of the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise stated herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each of the above embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0180] Based on the same inventive concept, the present application also provides a high-altitude operation safety detection device for implementing the above-mentioned high-altitude operation safety detection method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more high-altitude operation safety detection device embodiments provided below can refer to the limitations of the high-altitude operation safety detection method in the above text, which will not be repeated here.

[0181] In some embodiments, as shown in Figure 7 The present application also provides a high-altitude operation safety detection device, which comprises:

[0182] The acquisition module 10 is configured to acquire a to-be-detected image of a high-altitude operation area.

[0183] The first detection module 20 is configured to input the to-be-detected image into a visual detection model for detection to obtain a personnel detection result, wherein the detection result is used to determine whether the work personnel exists in the to-be-detected image.

[0184] The second detection module 30 is configured to, in a case where the work personnel exists in the to-be-detected image, detect the safety equipment of the work personnel in the to-be-detected image by using a first detection algorithm and / or a second detection algorithm to obtain a safety detection result.

[0185] In some embodiments, the device comprises:

[0186] The output module is configured to output a warning information in a case where any of the safety detection results indicates that the aerial work of the work personnel is unsafe, wherein the warning information is used to remind a user to check the aerial work safety in time.

[0187] In some embodiments, the first detection algorithm is a target detection model, and the second detection module 30 is configured to input the to-be-detected image into the target detection model to obtain the safety detection result, wherein a structure of the target detection model comprises an input end, a backbone network, a neck network and a prediction end.

[0188] The input end is configured to obtain a sample data set, perform mosaic data enhancement on a sample image in the sample data set, perform adaptive anchor frame calculation on the sample data set to determine an anchor frame size, and adaptively scale the sample image to a predetermined size.

[0189] The backbone network is configured to extract features of the sample image layer by layer.

[0190] The neck network is configured to perform feature enhancement and feature fusion according to a feature pyramid and a gate convolution.

[0191] The prediction end is configured to output the safety detection result corresponding to the sample image.

[0192] In some embodiments, the second detection model 30 comprises:

[0193] The first determination unit is configured to determine a first safety parameter corresponding to the safety equipment according to current use state information of the safety equipment.

[0194] The second determination unit is configured to determine a second safety parameter corresponding to a load-bearing frame of the work personnel according to current use information of the load-bearing frame.

[0195] The third determination unit is configured to determine the safety detection result according to the first safety parameter and / or the second safety parameter.

[0196] In some embodiments, the safety equipment comprises a safety helmet; the current usage state information comprises at least one of shell dent information, shell breakage information, and suspension system usage information; the shell dent information comprises dent degree and dent number of the safety helmet; the shell breakage information comprises breakage degree and breakage number of the safety helmet; the suspension system usage information comprises headband wear degree and suspension band wear degree.

[0197] In some embodiments, the safety equipment comprises a safety belt; the current usage state information comprises at least one of waist belt usage information, shoulder belt usage information, and back belt usage information; the waist belt usage information comprises waist belt breakage degree; the shoulder belt usage information comprises shoulder belt breakage degree; the back belt usage information comprises back belt breakage degree.

[0198] In some embodiments, the safety equipment comprises a safety glove; the current usage state information comprises at least one of glove wear information and hole information; the glove wear information comprises at least one of glove wear area and glove wear degree; the hole information comprises at least one of hole number and hole area.

[0199] In some embodiments, the second determining unit is configured to perform the following steps:

[0200] determining the first alternative parameter according to at least one of theoretical service life, remaining service life, rust number, rust area, deformation degree, crack degree, and crack number of the load-bearing frame;

[0201] determining the second alternative parameter according to the weight, number of the workers, and maximum load-bearing value of the load-bearing frame;

[0202] determining the second safety parameter according to the first alternative parameter and / or the second alternative parameter.

[0203] In some embodiments, the third determining unit is configured to perform one of the following:

[0204] Firstly, determining the safety detection result according to the first safety parameter;

[0205] Secondly, determining the safety detection result according to the second safety parameter;

[0206] Thirdly, determining the safety detection result according to sum, mean, ratio, and / or variance of the first safety parameter and the second safety parameter;

[0207] Fourthly, determining the safety detection result according to the product of the first safety parameter and a first influence factor, and the product of the second safety parameter and a second influence factor.

[0208] The modules in the above high-altitude operation safety detection device can be implemented by software, hardware, or a combination thereof. The modules can be embedded in or independent of a processor in the electronic device in hardware form, or stored in a memory in the electronic device in software form, so that the processor can call and execute the operations corresponding to the modules.

[0209] In one embodiment, an electronic device, which can be a terminal, has an internal structure as shown in Figure 8 The electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved by WIFI, mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program is executed by the processor to implement a high-altitude operation safety detection method. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad arranged on the housing of the electronic device, or an external keyboard, touchpad, or mouse, etc.

[0210] Those skilled in the art can understand that Figure 8 The structure shown in the above embodiment is only a block diagram of part of the structure related to the scheme of the present application, and does not limit the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0211] In one embodiment, an electronic device is provided, which includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps of the computer program implement the steps of the above method embodiments.

[0212] In one embodiment, a computer readable storage medium is provided, which stores a computer program. When the processor executes the computer program, the steps of the computer program implement the steps of the above method embodiments.

[0213] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties.

[0214] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments. Any reference to memory, database or other medium used in the embodiments provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0215] The technical features of the above embodiments can be combined in any way. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0216] The above-described embodiments are merely illustrative of several embodiments of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A method for detecting aerial work safety, characterized in that, The method comprises: obtaining a to-be-detected image of a high-altitude operation area; inputting the to-be-detected image into a visual detection model for detection to obtain a personnel detection result; wherein the detection result is used to determine whether an operation personnel exists in the to-be-detected image; in the case that the operation personnel exists in the to-be-detected image, detecting safety equipment of the operation personnel in the to-be-detected image through a first detection algorithm and / or a second detection algorithm respectively to obtain a safety detection result.

2. The method of claim 1, wherein, The method further comprises: in the case that any of the safety detection results indicates that high-altitude operation of the operation personnel is unsafe, outputting a warning information; wherein the warning information is used to remind a user to check high-altitude operation safety in time.

3. The method of claim 1, wherein, The first detection algorithm is a target detection model; the detection of the safety equipment of the operation personnel in the to-be-detected image through the first detection algorithm to obtain a safety detection result comprises: inputting the to-be-detected image into the target detection model to obtain the safety detection result; wherein the structure of the target detection model comprises an input end, a backbone network, a neck network and a prediction end: the input end is used to obtain a sample data set, perform mosaic data enhancement on a sample image in the sample data set; perform adaptive anchor box calculation on the sample data set to determine anchor box size, and adaptively scale the sample image to a predetermined size; the backbone network is used to extract features of the sample image layer by layer; the neck network is used to perform feature enhancement and feature fusion according to a feature pyramid and a gate convolution; the prediction end is used to output the safety detection result corresponding to the sample image.

4. The method of claim 1, wherein, The detection of the safety equipment of the operation personnel in the to-be-detected image through the second detection algorithm to obtain a safety detection result comprises: determining a first safety parameter corresponding to the safety equipment according to current use state information of the safety equipment; determining a second safety parameter corresponding to a load-bearing frame of the operation personnel according to current use information of the load-bearing frame; determining the safety detection result according to the first safety parameter and / or the second safety parameter.

5. The method of claim 4, wherein, The safety equipment comprises a safety helmet; the current use state information comprises at least one of shell indentation information, shell damage information and suspension system use information; the shell indentation information comprises indentation degree and indentation number of the safety helmet; the shell damage information comprises rupture degree and rupture number of the safety helmet; and the suspension system use information comprises headband wear degree and suspension belt wear degree.

6. The method of claim 4, wherein, The safety equipment comprises a safety belt; the current use state information comprises at least one of waistband use information, shoulder belt use information and back belt use information; the waistband use information comprises waistband rupture degree; the shoulder belt use information comprises shoulder belt rupture degree; and the back belt use information comprises back belt rupture degree.

7. The method of claim 4, wherein, The safety device includes a safety glove; the current use state information includes at least one of glove wear information and hole information; the glove wear information includes at least one of a glove wear area and a glove wear degree; and the hole information includes at least one of a number of holes and a hole area.

8. The method of claim 4, wherein, The second safety parameter corresponding to the load-bearing frame of the worker is determined according to the current use information of the load-bearing frame, and the method comprises the steps of: determining a first alternative parameter according to at least one of a theoretical service life, a remaining service life, a number of rusts, a rust area, a deformation degree, a cracking degree and a number of cracks of the load-bearing frame; determining a second alternative parameter according to the weight of the worker, the number and the maximum load-bearing value of the load-bearing frame; determining the second safety parameter according to the first alternative parameter and / or the second alternative parameter.

9. The method of claim 4, wherein, The safety detection result is determined according to the first safety parameter and / or the second safety parameter, and the method comprises one of the following: firstly, the safety detection result is determined according to the first safety parameter; secondly, the safety detection result is determined according to the second safety parameter; thirdly, the safety detection result is determined according to a sum, an average, a ratio and / or a variance of the first safety parameter and the second safety parameter; fourthly, the safety detection result is determined according to a product of the first safety parameter and a first influence factor and a product of the second safety parameter and a second influence factor.

10. A high altitude work safety detection device, characterized in that, The device comprises: an acquisition module configured to acquire a to-be-detected image of a high-altitude work area; a first detection module configured to input the to-be-detected image into a visual detection model for detection to obtain a personnel detection result, wherein the detection result is used to determine whether a worker exists in the to-be-detected image; a second detection module configured to, when the worker exists in the to-be-detected image, detect a safety device of the worker in the to-be-detected image through a first detection algorithm and / or a second detection algorithm to obtain a safety detection result.

Images