An intelligent safety helmet construction environment detection method and system based on image recognition

Abstract

The present invention relates to the field of image recognition technology, and in particular to an intelligent safety helmet construction environment detection method and system based on image recognition. The method includes obtaining image data of the construction environment; preprocessing the obtained image data, including denoising the construction image and enhancing the denoised image data; using the preprocessed image as input to construct a prediction model and introducing an attention mechanism for the prediction model; training the model according to the constructed prediction model, extracting features based on the trained prediction model, and judging the danger threshold according to the output image features. By constructing a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, the present invention can make full use of the advantages of different networks, thereby improving the judgment of dangerous situations.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an intelligent safety helmet construction environment detection method and system based on image recognition. Background Art

[0002] In various construction scenarios, the safety helmet, as an essential basic safety protection equipment for construction workers, plays a crucial role. It can effectively protect the heads of construction workers from object strikes and other injuries in case of accidents. However, the functions of traditional safety helmets are relatively single, only providing physical protection, lacking the ability to actively detect and warn of potential dangers in the construction environment. Traditional safety helmets mainly focus on the physical protection of the head and cannot monitor the construction environment in real time. When construction workers wear traditional safety helmets, they often have to rely on their own observations and experience to judge the safety status of the surrounding environment, which is prone to misjudgment or omission of potential dangers in complex construction environments.

[0003] In addition, traditional safety helmets lack the combination with modern technology and cannot achieve automated environment detection and warning. During the construction process, construction workers may ignore changes in the surrounding environment due to being focused on their work, resulting in the inability to react in time when a danger occurs. Some construction sites use sensors for environment detection, such as temperature sensors, humidity sensors, gas sensors, etc. Although sensors can monitor specific physical parameters in real time, their detection ranges are limited, often only able to detect a single type of risk factor, and cannot accurately judge following the specific working positions of construction workers. Moreover, sensors have weak object recognition and risk judgment capabilities in complex construction environments and cannot comprehensively evaluate the safety status of the construction environment. At present, there is a need for an intelligent safety helmet construction environment detection method and system based on image recognition. Summary of the Invention

[0004] To solve the problem that traditional safety helmets have weak object recognition and risk judgment capabilities in complex construction environments, the present invention provides an intelligent safety helmet construction environment detection method and system based on image recognition. The present invention combines image recognition technology with safety helmets, which can accurately identify risk factors in the construction environment and improve the accuracy and efficiency of detection.

[0005] In the first aspect, an intelligent safety helmet construction environment detection method based on image recognition provided by the present invention adopts the following technical solutions:

[0006] An intelligent safety helmet construction environment detection method based on image recognition includes:

[0007] Obtain image data of the construction environment;

[0008] Preprocess the acquired image data, including denoising the construction images and enhancing the denoised image data;

[0009] Use the preprocessed image as input to build a prediction model, including building a prediction model based on convolutional neural network, recurrent neural network and generative adversarial network, and introducing an attention mechanism for the prediction model;

[0010] Perform model training according to the built prediction model, including adjusting the parameters of the prediction model using an adaptive optimization algorithm;

[0011] Extract features based on the trained prediction model, including performing convolution operations on the input image with convolutional kernels to extract local features of the image;

[0012] Judge the danger threshold according to the output image features.

[0013] Further, the denoising process of the construction images and the enhancement of the denoised image data include replacing the gray value of each pixel point in the image data with the median value of the gray values in the neighborhood of the pixel point using median filtering, then calculating the gray histogram of the original image through histogram equalization, calculating the cumulative distribution function according to the gray histogram, and mapping the gray values of the original image through the cumulative distribution function to obtain the enhanced image.

[0014] Further, the construction of the prediction model based on convolutional neural network, recurrent neural network and generative adversarial network includes building a convolutional neural network composed of multiple convolutional layers, activation function layers, pooling layers and fully connected layers. Among them, the convolutional layer is composed of multiple convolutional kernels of different sizes. After the input image data passes through multiple convolutional layers, a set of feature maps is obtained. The different feature maps output by the convolutional layer are input to the activation function layer for non-linear transformation, and then enter the pooling layer for downsampling operation. Finally, the fully connected layer performs feature integration.

[0015] Further, the construction of the prediction model based on convolutional neural network, recurrent neural network and generative adversarial network also includes that the recurrent neural network uses a long short-term memory network. The input gate of the recurrent neural network is used to receive the feature information output by the fully connected layer. The output gate of the recurrent neural network transmits the output image data information to the generator of the generative adversarial network. The realistic construction environment image sequence generated by the adversarial network is used as additional training data and input into the recurrent neural network for training.

[0016] Further, introducing an attention mechanism to the prediction model includes obtaining two feature maps from the input feature map through two convolutional layers respectively, then performing element-wise multiplication and sigmoid activation on these two feature maps to obtain a spatial attention map. Finally, performing element-wise multiplication on the input feature map and the spatial attention map to obtain an enhanced feature map.

[0017] Further, adjusting the parameters of the model using an adaptive optimization algorithm includes calculating the exponential moving average of the squared model gradients for each parameter in the prediction model and updating the model parameters using the exponential moving average. The formula for the exponential moving average is:

[0018] ,

[0019] where, represents the exponential moving average of the squared gradient at time t, represents the decay rate, which is a value between 0 and 1, represents the exponential moving average of the squared gradient at the previous time, represents the gradient at the current time, that is, the derivative of the loss function with respect to the current parameter.

[0020] Further, judging the danger threshold according to the output image features includes classifying the image feature results output by the prediction model into dangerous equipment features, personnel behavior features, and environmental condition features, assigning weights to different features to construct a danger assessment model, and using the output structure of the danger assessment model to judge the danger threshold.

[0021] In a second aspect, an intelligent safety helmet construction environment detection system based on image recognition includes:

[0022] A data acquisition module configured to: acquire image data of the construction environment;

[0023] A preprocessing module configured to: preprocess the acquired image data, including denoising the construction image and enhancing the denoised image data;

[0024] A conversion module configured to: use the preprocessed image as input to construct a prediction model, including constructing a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, and introducing an attention mechanism to the prediction model;

[0025] A model module configured to: train the model according to the constructed prediction model, including adjusting the parameters of the prediction model using an adaptive optimization algorithm;

[0026] The feature extraction module is configured to: perform feature extraction based on the trained prediction model, including performing a convolution operation on the input image through a convolution kernel to extract local features of the image;

[0027] The transformation module is configured to: judge the danger threshold according to the output image features.

[0028] In a third aspect, the present invention provides a computer-readable storage medium, in which multiple instructions are stored, and the instructions are adapted to be loaded and executed by a processor of a terminal device for the intelligent safety helmet construction environment detection method based on image recognition.

[0029] In a fourth aspect, the present invention provides a terminal device, including a processor and a computer-readable storage medium. The processor is used to implement each instruction; the computer-readable storage medium is used to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor for the intelligent safety helmet construction environment detection method based on image recognition.

[0030] In summary, the present invention has the following beneficial technical effects:

[0031] 1. By constructing a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, the present invention can make full use of the advantages of different networks. The convolutional neural network extracts local features of the image and captures information such as the shape and texture of objects; the recurrent neural network processes the image sequence, records the dynamic changes during the construction process, and better identifies the development trend of potential dangers.

[0032] 2. By introducing an attention mechanism, the present invention further enhances the model's attention to key regions, improves the accuracy of feature extraction. For the input feature map, two feature maps are obtained through two convolutional layers and processed to obtain a spatial attention map, which is then multiplied element-wise with the input feature map, enabling the model to pay more attention to important regions, thereby improving the recognition accuracy of dangerous situations.

[0033] 3. The present invention uses an adaptive optimization algorithm to adjust the parameters of the prediction model. When the gradient is large, the learning rate is automatically adjusted to a larger value to approach the optimal solution faster. When the gradient is small, the learning rate is decreased to avoid oscillation near the optimal solution. This enables the model to be efficiently trained in different construction environments and tasks, improving the generalization ability of the model.

[0034] 4. The present invention judges the danger threshold according to the output image features, classifies the image feature results into dangerous equipment features, personnel behavior features, and environmental condition features, and constructs a danger assessment model by assigning weights to different features. This multi-factor comprehensive judgment method can quickly and accurately evaluate the danger of the construction environment. Description of the Drawings

[0035] Figure 1 It is a schematic diagram of the overall process of an intelligent safety helmet construction environment detection method based on image recognition according to an embodiment of the present invention. Specific implementation mode

[0036] The present invention will be further described in detail below with reference to the accompanying drawings.

[0037] Embodiment 1

[0038] Refer to Figure 1 , an intelligent safety helmet construction environment detection method based on image recognition in this embodiment includes:

[0039] Obtain image data of the construction environment;

[0040] Preprocess the obtained image data, including denoising the construction image and enhancing the denoised image data;

[0041] Use the preprocessed image as input to construct a prediction model, including constructing a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, and introducing an attention mechanism for the prediction model;

[0042] Train the prediction model according to the constructed prediction model, including adjusting the parameters of the prediction model using an adaptive optimization algorithm;

[0043] Extract features based on the trained prediction model, including performing a convolution operation on the input image with a convolution kernel to extract local features of the image;

[0044] Judge the danger threshold according to the output image features.

[0045] Specifically, an intelligent safety helmet construction environment detection method based on image recognition includes the following steps:

[0046] As Figure 1 shown, S1, obtain image data of the construction environment;

[0047] Select a high-resolution camera and install it on the intelligent safety helmet to ensure that the image of the construction environment can be clearly captured. Before installing the camera, it is necessary to analyze the overall layout of the construction site to understand the positions and importance of different construction stages and different operation areas. For example, for key parts such as high-altitude operation areas, large equipment operation areas, and hazardous material storage areas, it should be ensured that the camera can be directly aimed at or have a good view coverage.

[0048] S2, preprocess the obtained image data, including denoising the construction image and enhancing the denoised image data;

[0049] Median filtering is used to denoise the construction images. For a construction image, let the gray value of its pixel points be , where represents the coordinates of the pixel point. Determine a neighborhood size, for example, a square window of or . For each pixel point in the image, with this pixel point as the center, select all pixel points within its neighborhood, sort the gray values of the pixels in the neighborhood, and take the median as the new gray value of the central pixel point. The calculation formula is:

[0050] ,

[0051] where is the gray value of the original image at the coordinate , is the gray value of the filtered image at the coordinate , W is the image window, and represents the median operation.

[0052] After filtering, histogram equalization is used to enhance the denoised image.

[0053] First, calculate the gray histogram of the original image, where i represents the gray level, represents the number of pixel points with gray value i in the image. Then, calculate the cumulative distribution function ,

[0054] ,

[0055] where j represents the gray level, represents the cumulative number of pixel points with gray value less than or equal to i in the image. Finally, map the gray values of the original image through the cumulative distribution function to obtain the enhanced image. Let the gray value of the pixel point in the original image be , and the enhanced gray value is:

[0056] ,

[0057] where N is the total number of pixel points in the image and L is the total number of gray levels.

[0058] S3. Use the preprocessed image as input to construct a prediction model, including constructing a prediction model based on convolutional neural network, recurrent neural network, and generative adversarial network, and introducing an attention mechanism for the prediction model;

[0059] A Convolutional Neural Network (CNN) consists of an input layer, a convolutional layer, an activation function layer, a pooling layer, and a fully connected layer. Among them, the input layer receives the preprocessed construction environment image as input. The image is represented in the form of a matrix, where each element represents the color value or grayscale value of a pixel. The convolutional layer uses three convolutional kernels of different scales, namely , and . For the input image , after convolution operation, feature maps of different scales are obtained. For the convolutional kernel, the output feature map is , where and are the height and width of the output feature map, and is the number of channels. The convolution operation is expressed as:

[0060] ,

[0061] where represents the convolution operation, represents the convolutional kernel, and represents the bias term. The convolution operation is to slide the convolutional kernel pixel by pixel on the input image and perform element-wise multiplication and summation operations at each position. After the convolutional layer outputs the feature map, it enters the activation function layer. Among them, the ReLU activation function is adopted. The ReLU function remains unchanged when the input is positive and outputs zero when the input is negative, thus introducing non-linearity. Then, the pooling layer is used to perform downsampling on the feature map to reduce the resolution of the feature map, reduce the amount of calculation and the number of parameters. The pooling operation adopts the max pooling operation, and the maximum value in each region of the feature map is selected as the output. Finally, the fully connected layer performs feature integration, which is used to integrate the features extracted by the previous layers and output the final classification or prediction result. Each neuron in the fully connected layer is connected to all neurons in the previous layer, and linear transformation is performed through the weight matrix and non-linear transformation of the activation function. The last fully connected layer is usually used as the output layer, and corresponding results are output according to the task.

[0062] The recurrent neural network uses a long short-term memory network (LSTM). First, the construction environment image is input into the CNN for feature extraction. After being processed by multiple convolutional layers, activation function layers, and pooling layers in the CNN, a set of feature maps are obtained. These feature maps have a high dimension and can be converted into a one-dimensional vector through a flattening operation. The flattened feature vector is used as the input for one time step of the input sequence of the LSTM. The feature maps sequentially enter the input gate, forget gate, cell state, and output gate of the LSTM. The input gate performs a non-linear transformation on the current input and the hidden state at the previous moment to obtain a value between 0 and 1, indicating the importance of the input information. Its calculation formula is:

[0063] ,

[0064] where, is the input at the current moment, is the hidden state at the previous moment, is the cell state at the previous moment, , and are weight matrices, is the bias term, is the sigmoid function. The forget gate determines which information will be forgotten. The forget gate performs a non-linear transformation on the current input and the hidden state at the previous moment to obtain a value between 0 and 1, indicating which information in the cell state at the previous moment needs to be forgotten. Its calculation formula is the same as that of the input gate. The cell state is used to store long-term memory. The cell state controls through the input gate and the forget gate, stores new information, and forgets old information, thereby updating the cell state. The calculation formula is:

[0065] ,

[0066] where, is the output of the forget gate, is the element-wise product, is the hyperbolic tangent function, is the bias term, and are weight matrices. Finally, the output is sent to the generative adversarial network (GAN) through the output gate. The GAN is mainly used to generate realistic construction environment image data, expand the training dataset, and improve the generalization ability of the model. The GAN consists of a generator and a discriminator. The goal of the generator is to generate fake images similar to real images to deceive the discriminator; the goal of the discriminator is to distinguish between real images and fake images generated by the generator.

[0067] When training a GAN, by alternately optimizing the parameters of the generator and the discriminator, the generator gradually generates more realistic images. Random noise can be used as the input of the generator, and fake images are generated through a multi-layer neural network. The discriminator receives real images and fake images as inputs and outputs a probability value indicating the probability that the input image is a real image. To sum up, the overall construction process of the prediction model is that real construction environment images are input into the CNN for feature extraction. The sequence of feature vectors output by the CNN is input into the LSTM. The LSTM learns the temporal dependencies of the image sequence. At the same time, the output of the LSTM can be fed back to the CNN for feature fusion optimization. On the other hand, the output of the LSTM can also guide the generator of the GAN to generate a new image sequence. The image sequence generated by the GAN can be input into the CNN and LSTM together with real images for training, forming a closed-loop system.

[0068] After the model is constructed, the attention mechanism is introduced. First, calculate the spatial attention map. For the input feature map , two feature maps and are obtained through two convolutional layers respectively. Then, these two feature maps are multiplied element-wise and sigmoid-activated to obtain the spatial attention map , that is, . Finally, the input feature map is multiplied element-wise with the spatial attention map to obtain the enhanced feature map .

[0069] S4. According to the constructed prediction model, perform model training, including using an adaptive optimization algorithm to adjust the parameters of the prediction model;

[0070] Set the initial parameter values, including the learning rate , the decay rate (between 0 and 1), the constant (used to prevent the denominator from being zero), etc. At the same time, initialize the exponential moving average of the squared gradient to a small positive number. For each parameter in the prediction model, calculate the derivative of the loss function with respect to this parameter, that is, the gradient at the current moment. Calculate the exponential moving average of the squared model gradient according to each parameter in the prediction model, and use the exponential moving average to update the model parameters. The calculation formula of the exponential moving average is:

[0071] ,

[0072] where represents the exponential moving average of the squared gradient at time t, Denoted as the attenuation rate, it is a value between 0 and 1. Denoted as the exponentially weighted moving average of the square of the gradient at the previous moment. Denoted as the gradient at the current moment, that is, the derivative of the loss function with respect to the current parameter.

[0073] ,

[0074] Among them, is the current parameter value, is the learning rate, is a constant used to prevent the denominator from being zero. For example, for the current parameter , the learning rate , the current gradient , the exponentially weighted moving average obtained from the previous calculation , then the updated parameter is:

[0075] ,

[0076] The new parameter value can be obtained through calculation.

[0077] S5. Feature extraction is performed based on the trained prediction model, including performing a convolution operation on the input image with a convolution kernel to extract local features of the image;

[0078] The construction image data obtained on-site is uploaded to the prediction model. The convolutional layer in the CNN performs a convolution operation on the input image with a convolution kernel to extract local features of the image. For an input image of size and a convolution kernel of size , the calculation formula for the convolution operation is:

[0079] ,

[0080] Among them, is the grayscale value of the input image at the coordinate , is the grayscale value of the image after convolution at the coordinate , is the weight of the convolution kernel at the coordinate , is the bias term. The pooling layer reduces the resolution of the image through downsampling the image after convolution, reduces the computational amount, and simultaneously extracts global features of the image.

[0081] In this embodiment, max pooling is adopted. For an input image of size and a pooling window of size , the calculation formula for max pooling is:

[0082] ,

[0083] Among them, is the grayscale value of the input image at the coordinate , is the grayscale value of the image after pooling at the coordinate , is the pooling window.

[0084] S6. Judge the danger threshold according to the output image features.

[0085] Based on the image feature results output by the prediction model, mark them as dangerous equipment features, personnel behavior features, and environmental condition features. Among them, dangerous equipment features: include the dangerous equipment existing in the construction site, as well as the shape, color, and signs of the equipment; personnel behavior features include the distance between construction workers and the dangerous area; environmental condition features include whether there are obstacles and dangerous areas (such as deep pits, edges of high-altitude operation platforms, etc.) in the current construction environment. For different features, weight distribution is carried out to construct a danger assessment model. Among them, the weight distribution uses the Analytic Hierarchy Process (AHP) to determine the weights of each feature. First, construct a judgment matrix to compare the relative importance between different features; then, determine the weights of each feature by calculating the eigenvector of the judgment matrix, and construct a danger assessment model using the linear weighted method. Among them, the expression of the danger assessment model is:

[0086] ,

[0087] Among them, represents the danger assessment value, , and respectively represent the values of dangerous equipment features, personnel behavior features, and environmental condition features, , and respectively represent the weights of each feature. Set different thresholds according to different types of construction projects. For example, bridge construction usually involves high-altitude operations and large machinery operations, with a relatively high danger level, and the danger threshold can be set to 65. Road construction is mainly ground operations, with a relatively low danger level, and the danger threshold can be set to 75. Compare the danger assessment value with the threshold, and give a danger prompt when it exceeds the threshold.

[0088] Embodiment 2

[0089] The difference between this embodiment and Embodiment 1 is that this embodiment provides an intelligent safety helmet construction environment detection system based on image recognition, including:

[0090] A data acquisition module, configured to: acquire image data of the construction environment;

[0091] A preprocessing module, configured to: preprocess the acquired image data, including denoising the construction image and enhancing the denoised image data;

[0092] A conversion module, configured to: use the preprocessed image as input to construct a prediction model, including constructing a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, and introducing an attention mechanism for the prediction model;

[0093] A model module, configured to: perform model training according to the constructed prediction model, including adjusting the parameters of the prediction model using an adaptive optimization algorithm;

[0094] A feature extraction module, configured to: extract features based on the trained prediction model, including performing a convolution operation on the input image with a convolution kernel to extract local features of the image;

[0095] A transformation module, configured to: judge a danger threshold according to the output image features.

[0096] A computer-readable storage medium, in which multiple instructions are stored, and the instructions are adapted to be loaded and executed by a processor of a terminal device for an intelligent safety helmet construction environment detection method based on image recognition as described above.

[0097] A terminal device, including a processor and a computer-readable storage medium, where the processor is used to implement each instruction; the computer-readable storage medium is used to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor for an intelligent safety helmet construction environment detection method based on image recognition as described above.

[0098] The above are all preferred embodiments of the present invention, and the protection scope of the present invention is not limited accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of the present invention shall be covered by the protection scope of the present invention.

Claims

1. A method for detecting construction environment of intelligent safety helmets based on image recognition, characterized in that: include: Acquire image data of the construction environment; Preprocessing the acquired image data, including denoising the construction image and enhancing the denoised image data; Using the preprocessed image as input to build a prediction model, including building a prediction model based on a convolutional neural network, a recurrent neural network and a generative adversarial network, and introducing an attention mechanism for the prediction model, wherein the prediction model based on the convolutional neural network, the recurrent neural network and the generative adversarial network is built, including the recurrent neural network using a long short-term memory network, the input gate of the recurrent neural network is used to receive feature information output by a fully connected layer, the output gate of the recurrent neural network transmits output image data information to the generator of the generative adversarial network, and the realistic construction environment image sequence generated by the adversarial network is used as additional training data and input into the recurrent neural network for training; Performing model training based on the constructed prediction model, including adjusting the parameters of the prediction model using an adaptive optimization algorithm; Perform feature extraction based on the trained prediction model, including performing convolution operations with the input image through the convolution kernel to extract local features of the image; The hazard threshold is judged according to the output image features, including dividing the image feature results output by the prediction model into dangerous equipment features, personnel behavior features and environmental condition features, and constructing a hazard assessment model by weight allocation for different features. The hazard threshold is judged using the output structure of the hazard assessment model, wherein the weight allocation adopts the analytic hierarchy process (AHP) to determine the weight of each feature. First, a judgment matrix is ​​constructed to compare the relative importance of different features. Then, the weight of each feature is determined by calculating the eigenvector of the judgment matrix, and the hazard assessment model is constructed using the linear weighted method.

2. According to the method for detecting the construction environment of an intelligent helmet based on image recognition according to claim 1, it is characterized in that: The construction image is denoised and the denoised image data is enhanced, including using a median filter to replace the grayscale value of each pixel in the image data with the median of the grayscale values ​​in the neighborhood of the pixel, then calculating the grayscale histogram of the original image through histogram equalization, calculating the cumulative distribution function based on the grayscale histogram, and mapping the grayscale values ​​of the original image through the cumulative distribution function to obtain an enhanced image.

3. The method for detecting construction environment of an intelligent safety helmet based on image recognition according to claim 2 is characterized in that: The method comprises constructing a prediction model based on a convolutional neural network, a recurrent neural network and a generative adversarial network, comprising constructing a convolutional neural network composed of a plurality of convolutional layers, an activation function layer, a pooling layer and a fully connected layer, wherein the convolutional layer is composed of a plurality of convolutional kernels of different sizes, and the input image data is passed through a plurality of convolutional layers to obtain a set of feature maps, and the different feature maps output by the convolutional layer are input to the activation function layer for nonlinear transformation, and then enter the pooling layer for downsampling operation, and finally the fully connected layer performs feature integration.

4. The method for detecting construction environment of an intelligent safety helmet based on image recognition according to claim 3 is characterized in that: The attention mechanism is introduced into the prediction model, including obtaining two feature maps through two convolutional layers for the input feature map respectively, and then performing element-by-element multiplication and sigmoid activation on the two feature maps to obtain a spatial attention map, and finally, performing element-by-element multiplication of the input feature map and the spatial attention map to obtain an enhanced feature map.

5. The method for detecting construction environment of an intelligent helmet based on image recognition according to claim 4 is characterized in that: The method of adjusting the parameters of the model using the adaptive optimization algorithm includes calculating the exponential moving average of the square of the model gradient according to each parameter in the prediction model, and updating the model parameters using the exponential moving average. The calculation formula of the exponential moving average is: ， in, It is expressed as the exponential moving average of the square of the gradient at time t, It is expressed as a decay rate, which is a value between 0 and 1. It is expressed as the exponential moving average of the square of the gradient at the previous moment, It is expressed as the gradient at the current moment, that is, the derivative of the loss function with respect to the current parameters.

6. An intelligent safety helmet construction environment detection system based on image recognition, executing the method according to any one of claims 1 to 5, characterized in that: include: The data acquisition module is configured to: acquire image data of the construction environment; The preprocessing module is configured to: preprocess the acquired image data, including denoising the construction image, and performing image enhancement on the denoised image data; The conversion module is configured to: use the preprocessed image as input to build a prediction model, including building a prediction model based on a convolutional neural network, a recurrent neural network, and a generative adversarial network, and introduce an attention mechanism for the prediction model; The model module is configured to: perform model training according to the constructed prediction model, including adjusting the parameters of the prediction model by using an adaptive optimization algorithm; The feature extraction module is configured to: perform feature extraction based on the trained prediction model, including performing a convolution operation with the input image through a convolution kernel to extract local features of the image; The transformation module is configured to: determine the danger threshold according to the output image features.

7. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instructions are suitable for being loaded and executed by a processor of a terminal device according to a method for detecting a construction environment of an intelligent safety helmet based on image recognition as claimed in claim 1.

8. A terminal device, comprising a processor and a computer-readable storage medium, wherein the processor is used to implement each instruction; and the computer-readable storage medium is used to store multiple instructions, characterized in that: The instructions are suitable for being loaded by a processor and executing the method for detecting construction environment of an intelligent safety helmet based on image recognition as claimed in claim 1.

Images