Intelligent identification and early warning method and device for high and steep slope heap

Abstract

The invention relates to the technical field of monitoring and early warning, and discloses an intelligent identification and early warning method and device for a high and steep slope heap, and the method comprises the steps: constructing a high and steep slope heap sample library, and carrying out the adjustment of a first parameter of an initial score function based on the high and steep slope heap sample library, and obtaining a target score function containing a second parameter; obtaining stacked object images, and classifying the stacked object images based on the target score function to obtain category probabilities; inputting the category probability and the heap image into a preset neural network model, and determining a heap type and a coverage area proportion; and determining a classification result based on the heap type and the coverage area proportion, and executing an early warning measure corresponding to the classification result. According to the method, the high and steep slope heap sample library is constructed and the initial score function is adjusted, so that the features of the heap are reflected more accurately, the classification accuracy is improved, and the recognition precision of the heap type and the coverage area ratio is further improved in combination with the preset neural network model.

Description

Technical Field

[0001] This application relates to the technical field of monitoring and early warning, and particularly to an intelligent identification and early warning method and device for piled objects on high and steep slopes. Background Art

[0002] The intelligent identification and early warning system for piled objects on high and steep slopes is an important safety monitoring system. In the construction section, some operators, for the convenience of construction, temporarily pile building materials and waste on the nearby high and steep slopes, which is extremely likely to cause slope instability, resulting in safety accidents and economic losses. Therefore, in view of the above situation, an efficient identification and early warning system needs to be designed. This system should be able to timely and accurately identify whether there are piled objects on the slope, how much the piled volume is, and transmit the early warning signal to on-site technical personnel and remote supervision personnel in time to take relevant measures to eliminate risks in advance.

[0003] The existing solutions mainly determine the stability of the slope by comparing the displacement change amount of the monitoring point with the early warning value. It has a certain effect on the monitoring of ordinary slope piled objects, but for the situation of high and steep slope instability, due to the large instantaneous change amount of displacement when the high and steep slope is unstable, the early warning effect of this solution is lagging and not suitable for the monitoring of high and steep slopes. Summary of the Invention

[0004] To solve the problem of the large instantaneous change amount of displacement when the high and steep slope is unstable mentioned above, the embodiments of this application provide an intelligent identification and early warning method and device for piled objects on high and steep slopes, and its technical solutions are as follows: In a first aspect, the embodiments of this application provide an intelligent identification and early warning method for piled objects on high and steep slopes, including: Construct a sample library of piled objects on high and steep slopes, and adjust the first parameter of the initial scoring function based on the sample library of piled objects on high and steep slopes to obtain a target scoring function including a second parameter; Obtain a piled object image, and classify the piled object image based on the target scoring function to obtain a category probability; Input the category probability and the piled object image into a preset neural network model to determine the piled object type and the proportion of the covered area; Determine a classification result based on the piled object type and the proportion of the covered area, and execute an early warning measure corresponding to the classification result.

[0005] In an alternative solution of the first aspect, constructing a sample library of piled objects on high and steep slopes includes: Collect sample piled object images of high and steep slopes through a camera device, and perform annotation processing on the sample piled object images to obtain a sample library including at least five piled object types.

[0006] In yet another alternative of the first aspect, classifying the heap image based on the target scoring function to obtain class probabilities, including: Extracting the feature vector of the heap image and inputting the feature vector into the target scoring function to calculate the class probabilities of at least five heap types corresponding to the heap image.

[0007] In yet another alternative of the first aspect, inputting the class probabilities and the heap image into a preset neural network model to determine the heap type and the proportion of the covered area, including: Extracting features of the heap image through the neural network model and matching the extracted features with the preset heap type features to determine the heap type; Segmenting the heap image through the neural network model and calculating the ratio of the heap area to the entire slope area to obtain the proportion of the covered area.

[0008] In yet another alternative of the first aspect, determining the classification result based on the heap type and the proportion of the covered area and performing the warning measures corresponding to the classification result, including: When it is detected that the proportion of the covered area is greater than or equal to a preset first ratio, calculating the distance to the slope edge; When it is detected that the distance to the slope edge is less than or equal to the preset distance, triggering an orange warning light and sending a text message notification to the management personnel.

[0009] In yet another alternative of the first aspect, determining the classification result based on the heap type and the proportion of the covered area and performing the warning measures corresponding to the classification result, further including: When it is detected that the proportion of the covered area is greater than or equal to a preset second ratio, triggering a red warning light and generating an emergency work order.

[0010] In yet another alternative of the first aspect, after performing the warning measures corresponding to the classification result, further including: Reconstructing the heap image through an autoencoder and calculating the pixel-level mean square error of at least two regions; If there is a continuous region where the pixel-level mean square error is greater than or equal to the preset threshold, marking the heap image as invalid data.

[0011] In a second aspect, an intelligent recognition and warning device for high-steep slope heap objects provided by an embodiment of the present application includes: A first processing module, configured to construct a high-steep slope heap object sample library and adjust the first parameter of the initial scoring function based on the high-steep slope heap object sample library to obtain a target scoring function including a second parameter; A second processing module, configured to obtain a heap image and classify the heap image based on the target scoring function to obtain class probabilities; The third processing module is configured to input the category probability and the piled object image into a preset neural network model to determine the type of the piled object and the proportion of the covered area; The fourth processing module is configured to determine a classification result based on the type of the piled object and the proportion of the covered area, and execute a warning measure corresponding to the classification result.

[0012] In a third aspect, an embodiment of the present application further provides an intelligent recognition and warning device for piled objects on a high-steep slope, including a processor and a memory; The processor is connected to the memory; The memory is configured to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the intelligent recognition and warning method for piled objects on a high-steep slope provided in the first aspect or any implementation manner of the first aspect of the embodiments of the present application.

[0013] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, the intelligent recognition and warning method for piled objects on a high-steep slope provided in the first aspect or any implementation manner of the first aspect of the embodiments of the present application can be implemented.

[0014] The beneficial effects brought by the technical solutions provided in some embodiments of this specification at least include: In the process of intelligent recognition and warning of piled objects on a high-steep slope, by constructing a sample library of piled objects on a high-steep slope and adjusting the initial score function, the characteristics of the piled objects can be more accurately reflected, thereby improving the accuracy of classification. Furthermore, in combination with a preset neural network model, the recognition accuracy of the type of the piled object and the proportion of the covered area is further improved, realizing high-precision intelligent recognition and warning of piled objects on a high-steep slope. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is the overall flowchart of an intelligent recognition and warning method for piled objects on a high-steep slope provided by an embodiment of the present application; Figure 2 It is the structural schematic diagram of an intelligent recognition and warning device for piled objects on a high-steep slope provided by an embodiment of the present application; Figure 3This is a schematic structural diagram of another intelligent recognition and early warning device for piled objects on high and steep slopes provided by the embodiments of the present application. Detailed implementation manners

[0017] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application.

[0018] In the following introduction, the terms "first" and "second" are only for the purpose of description and cannot be construed as indicating or implying relative importance. The following introduction provides multiple embodiments of the present application. Different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing all other possible combinations of A, B, C, and D, even though such embodiments may not be explicitly described in the following content.

[0019] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes can be made to the functions and arrangements of the described elements without departing from the scope of the content of the present application. Each example can appropriately omit, substitute, or add various processes or components. For example, the described method can be executed in a different order from the described order, and various steps can be added, omitted, or combined. In addition, the features described in some examples can be combined into other examples.

[0020] Please refer to Figure 1 , Figure 1 which shows an overall flowchart of an intelligent recognition and early warning method for piled objects on high and steep slopes provided by the embodiments of the present application.

[0021] As Figure 1 shown, the intelligent recognition and early warning method for piled objects on high and steep slopes can at least include the following steps: Step 101: Construct a sample library of piled objects on high and steep slopes, and adjust the first parameter of the initial scoring function based on the sample library of piled objects on high and steep slopes to obtain a target scoring function containing the second parameter.

[0022] In the embodiments of the present application, the intelligent identification and early warning method for high-steep slope deposits can be but is not limited to being applied to personal computers, industrial computers, embedded devices, servers, cloud platforms, edge computing devices, robots, Internet of Things devices, FPGA devices, and high-performance computing clusters, etc. During the intelligent identification and early warning process of high-steep slope deposits, by constructing a sample library of high-steep slope deposits and adjusting the initial scoring function, the characteristics of the deposits can be more accurately reflected, thereby improving the accuracy of classification. Furthermore, combined with a preset neural network model, the identification accuracy of the deposit type and the proportion of the covered area is further improved, realizing high-precision intelligent identification and early warning of high-steep slope deposits.

[0023] Specifically, during the intelligent identification and early warning process of high-steep slope deposits, a sample library of high-steep slope deposits can be constructed first, that is, a large amount of image data of high-steep slope deposits is collected, covering different deposits (such as earth and stone, construction waste, etc.) and their performances under different environments (such as different lighting, angles, covered areas, etc.), and the image data is labeled to clarify information such as the type, boundary, and covered area of the deposits, forming a high-quality sample library (such as a slope dataset without deposits, a slope dataset with light objects covering a small area, a slope dataset with light objects covering a large area, a slope dataset with heavy objects covering a small area, a slope dataset with heavy objects covering a large area).

[0024] Furthermore, an initial scoring function such as f(x,W)=Wx + b can be constructed, where x is the size of each pixel point in the image, W is the proportion weight of each category, and b is the classification operation number of the image recognition result category. It can be understood that the initial scoring function is used to evaluate the matching degree between the characteristics of the deposit image and the target category. Its first parameter is usually set based on experience or preliminary data, and the first parameter can be but is not limited to including one or both of W and b. After determining the sample library of high-steep slope deposits and the initial scoring function, the first parameter of the initial scoring function can be optimized and adjusted through the data in the sample library, and then a target scoring function containing a second parameter can be obtained to make it more accurately reflect the characteristics of the deposits. For example, iteratively adjust W and b according to the error feedback until the classification accuracy rate ≥ 98%.

[0025] It should be noted that the second parameter is the adjusted first parameter, and the target scoring function can more comprehensively evaluate the characteristics of the deposits and improve the accuracy of classification.

[0026] Suppose in a high-steep slope area of a mine, 1000 stack images are taken by a drone, including types such as earth-rock stacks and construction waste stacks. These images are labeled, for example, the boundaries and covered areas of the earth-rock stacks are labeled, etc., to form a sample library. The initial scoring function may be evaluated based on the color characteristics of the stack, but its parameter settings are relatively simple, resulting in poor classification effects. Through the data in the sample library, it is found that the color characteristics of the earth-rock stacks vary greatly under different illuminations. Therefore, the first parameter (such as the color weight) is adjusted to make it more adaptable to the actual scenario. The adjusted color characteristic parameter (the first parameter) is used as the second parameter to further optimize the scoring function. For example, the target scoring function not only uses the adjusted color characteristic parameter, but also combines the texture and shape characteristics of the stack, so as to more accurately distinguish between earth-rock stacks and construction waste stacks.

[0027] As an option of the embodiment of the present application, constructing a high-steep slope stack sample library includes: Collecting sample stack images of the high-steep slope through a camera device, and performing labeling processing on the sample stack images to obtain a sample library containing at least five types of stacks.

[0028] Specifically, in the process of constructing the high-steep slope stack sample library, a high-definition camera device (such as a drone, a surveillance camera, etc.) can be used to collect images of the high-steep slope area first, ensuring that different angles, illumination conditions, and stack states are covered to obtain sample stack images. It should be noted that the collected images need to be representative and can reflect various stack types and their distribution situations that may appear in the actual scenario.

[0029] Then, the collected images can be manually or semi-automatically labeled to clearly label key information such as the type, boundary, and covered area of the stack. During the labeling process, the accuracy and consistency of the data need to be ensured to provide high-quality basic data for subsequent model training.

[0030] After that, a sample library can be constructed according to the results of the labeling process. Among them, the sample library needs to cover at least five types of stacks (such as a slope data set without stacks, a slope data set with light objects covering a small area, a slope data set with light objects covering a large area, a slope data set with heavy objects covering a small area, a slope data set with heavy objects covering a large area, etc.) to ensure the generalization ability of the model. It should be noted that the number of samples of each type of stack should be balanced to avoid data deviation affecting the model performance.

[0031] Step 102, obtaining a stack image, and classifying the stack image based on the target scoring function to obtain a class probability.

[0032] Specifically, after obtaining the target scoring function, stack images of the high-steep slope area can be collected in real time through a camera device (such as a drone, a surveillance camera, etc.), and it is ensured that the images are clear and cover the target area.

[0033] Next, the heap image can be input into the target scoring function to calculate the matching scores with various categories (such as small-area coverage of light objects, large-area coverage of heavy objects, etc.). Thus, according to the calculation results of the target scoring function, the scores can be converted into probability values through normalization processing (such as the Softmax function). The probability value of each category represents the possibility that the heap image belongs to that category. The higher the probability value, the higher the degree of matching.

[0034] As another alternative in the embodiments of the present application, classifying the heap image based on the target scoring function to obtain category probabilities includes: Extracting the feature vector of the heap image and inputting the feature vector into the target scoring function to calculate the category probabilities of at least five heap types corresponding to the heap image.

[0035] Specifically, after obtaining the heap image, a feature extraction method (such as a convolutional neural network CNN, traditional image processing algorithms, etc.) can be used to extract key features from the heap image to obtain a feature vector. Among them, the feature vector is a mathematical representation of the heap image in the feature space and can reflect information such as the color, texture, shape, and edge of the heap.

[0036] Next, the extracted feature vector can be input into the target scoring function to calculate its matching scores with various categories. Thus, according to the calculation results of the target scoring function, the scores can be converted into probability values through normalization processing (such as the Softmax function) to obtain the possibility that the heap image belongs to each category. The category with a higher probability value is more likely to be the category to which the heap image belongs.

[0037] Step 103: Input the category probabilities and the heap image into a preset neural network model to determine the heap type and the proportion of the covered area.

[0038] Specifically, after obtaining the category probabilities, the category probabilities of the heap image (such as 65% for small-area coverage of light objects, 20% for large-area coverage of heavy objects, etc.) and the original heap image can be used as inputs and passed into the preset neural network model.

[0039] It can be understood that the category probabilities provide a preliminary judgment of the heap type, while the original image retains the spatial information and detailed features of the heap. The preset neural network model is usually a multi-task learning model that can simultaneously process the tasks of heap type recognition and covered area calculation. The model may include a convolutional neural network (CNN) for feature extraction, and a fully connected layer or a segmentation network (such as U-Net) for type recognition and area calculation.

[0040] Next, the neural network model can further refine the judgment of the piled object type based on the input class probabilities and image features. For example, the model may combine the spatial information of the image to correct the preliminary class probabilities and finally determine the piled object type (such as light objects covering a small area). The model uses image segmentation techniques (such as semantic segmentation) to identify the specific area of the piled object in the image and calculate its coverage area. The coverage area ratio refers to the ratio of the piled object area to the entire image area. For example, the piled object covers 30% of the image area.

[0041] As another option in the embodiments of this application, inputting the class probabilities and the piled object image into a preset neural network model to determine the piled object type and the coverage area ratio includes: Extracting features from the piled object image through the neural network model and matching the extracted features with the preset piled object type features to determine the piled object type; Segmenting the piled object image through the neural network model and calculating the ratio of the piled object area to the entire slope area to obtain the coverage area ratio.

[0042] Specifically, in the process of determining the piled object type and the coverage area ratio based on the class probabilities and the piled object image, high-level features can be first extracted from the piled object image through a neural network model (such as a convolutional neural network CNN). The extracted features are a multi-dimensional vector used to represent the distribution of the piled object image in the feature space, and these features can capture key information such as the color, texture, shape, and edges of the piled object.

[0043] Next, the extracted feature vector can be compared with the preset piled object type features (such as earth and stone piles, construction waste piles, etc.) to calculate their similarity. Thus, in combination with the input class probabilities, the matching result can be further optimized through the neural network model to finally determine the piled object type.

[0044] After that, image segmentation techniques (such as the semantic segmentation network U-Net) can be used to process the piled object image to identify the piled object area and the background area. Thus, according to the segmentation result, the ratio of the number of pixels in the piled object area to the number of pixels in the entire slope area can be statistically calculated. It can be understood that the segmentation result is a pixel-level classification map clearly marking the specific position of the piled object in the image, and the coverage area ratio = (number of pixels in the piled object area / number of pixels in the entire slope area) × 100%.

[0045] Step 104: Determine the classification result based on the piled object type and the coverage area ratio, and execute the warning measures corresponding to the classification result.

[0046] Specifically, in the process of determining the classification result, the risk level of the piled objects, that is, the classification result, can be determined first according to the type of piled objects and the proportion of the covered area, in combination with the preset classification rules. Among them, the classification rules can include, but are not limited to: the risk coefficient corresponding to the type of piled objects (for example, the risk of earth and stone piles is relatively high, and the risk of vegetation piles is relatively low) and the threshold of the proportion of the covered area (for example, more than 30% is high risk, 10%-30% is medium risk, and less than 10% is low risk).

[0047] Furthermore, corresponding warning measures can be triggered according to the classification result. For example, high risk: issue an emergency alarm and notify relevant personnel to take measures immediately (such as evacuation, slope reinforcement, etc.); medium risk: issue a warning notice and recommend strengthening monitoring and taking preventive measures; low risk: record data and continuously monitor, without the need to take immediate measures.

[0048] As another option of the embodiment of the present application, determining the classification result based on the type of piled objects and the proportion of the covered area, and executing the warning measures corresponding to the classification result, includes: When it is detected that the proportion of the covered area is greater than or equal to the preset first ratio, calculate the distance to the slope edge; When it is detected that the distance to the slope edge is less than or equal to the preset distance, trigger an orange warning light and send a text message notification to the management personnel.

[0049] Specifically, in the process of determining the classification result and executing the warning measures corresponding to the classification result, it can be first detected whether the proportion of the covered area is greater than or equal to the preset first ratio. Among them, the preset first ratio is a threshold (for example, 30%) for judging whether the covered area of the piled objects reaches the medium-risk level. If the proportion of the covered area ≥ the first ratio, the system enters the next step of calculating the distance to the slope edge. Here, image processing technology or deep learning methods can be used to identify the closest distance between the piled object area and the slope edge, and then the spatial relationship between the boundary pixels of the piled object area and the slope edge pixels can be calculated based on the distance.

[0050] It can be understood that the preset distance is a safety threshold (for example, 5 meters) for judging whether the piled objects are too close to the slope edge. If the distance to the slope edge ≤ the preset distance, the system determines it as a medium-risk situation. At this time, an orange warning light can be used for on-site warning to remind relevant personnel to pay attention to the high-risk situation. At the same time, the system automatically sends a text message notification to the management personnel, including information such as the type of piled objects, the proportion of the covered area, and the distance to the slope edge, so as to take measures in time.

[0051] As another option of the embodiment of the present application, determining the classification result based on the type of piled objects and the proportion of the covered area, and executing the warning measures corresponding to the classification result, further includes: When the detected coverage area ratio is greater than or equal to a preset second ratio, trigger a red warning light and generate an emergency work order.

[0052] Specifically, during the process of determining the classification result and implementing the warning measures corresponding to the classification result, it is also possible to detect whether the coverage area ratio is greater than or equal to a preset second ratio. Among them, the preset second ratio is a higher threshold (such as 50%) used to determine whether the coverage area of the piled objects reaches a high-risk level. If the coverage area ratio ≥ the second ratio, the system determines it as a high-risk situation. At this time, a red warning light can be used for on-site emergency warning to prompt relevant personnel to take corresponding measures immediately. It can be understood that the triggering of the red warning light indicates that the situation is very urgent and there may be serious landslide or collapse risks.

[0053] Then, the system can automatically generate an emergency work order. The content of the work order can but is not limited to including the type of piled objects, the coverage area ratio, the risk level, and recommended measures (such as immediate evacuation, slope reinforcement, etc.). The emergency work order will be directly sent to the relevant responsible departments or personnel to ensure a quick response.

[0054] As another option in the embodiment of the present application, after implementing the warning measures corresponding to the classification result, it further includes: Reconstruct the piled object image through an autoencoder and calculate the pixel-level mean squared error of at least two regions; If there are consecutive regions where the pixel-level mean squared error is greater than or equal to a preset threshold, mark the piled object image as invalid data.

[0055] Specifically, after implementing the warning measures corresponding to the classification result, the piled object image can also be reconstructed through an autoencoder. It can be understood that an autoencoder is a neural network model composed of an encoder and a decoder, which can learn the low-dimensional representation of the input data and reconstruct the original data. Input the piled object image into the autoencoder, the encoder compresses it into a low-dimensional feature representation, and the decoder reconstructs the image based on this feature.

[0056] After reconstructing the piled object image, the reconstructed image can also be compared with the original piled object image to calculate the pixel-level mean squared error (MSE). Among them, at least two regions (such as the piled object region and the background region) can be selected to calculate the MSE respectively, and then it can be determined whether there are consecutive regions where the pixel-level MSE is greater than or equal to a preset threshold. It can be understood that the preset threshold is an empirical value used to determine whether the difference between the reconstructed image and the original image is within an acceptable range. If the MSE of a certain consecutive region ≥ the preset threshold, it means that the reconstruction effect of the autoencoder in this region is poor, which may be due to image quality problems or complex piled object features. Therefore, the piled object image can be marked as invalid data, and the invalid data will be excluded from the training or analysis process to ensure the accuracy and reliability of the model.

[0057] Please refer to Figure 2 , Figure 2 which shows a schematic structural diagram of an intelligent recognition and early warning device for high-steep slope stacking provided by an embodiment of the present application.

[0058] As Figure 2 shown, the intelligent recognition and early warning device for high-steep slope stacking may at least include a first processing module 201, a second processing module 202, a third processing module 203, and a fourth processing module 204, where:[[]] The first processing module 201 is used to construct a high-steep slope stacking sample library and adjust the first parameter of the initial scoring function based on the high-steep slope stacking sample library to obtain a target scoring function including a second parameter; The second processing module 202 is used to acquire a stacking image and classify the stacking image based on the target scoring function to obtain a category probability; The third processing module 203 is used to input the category probability and the stacking image into a preset neural network model to determine the stacking type and the proportion of the covered area; The fourth processing module 204 is used to determine a classification result based on the stacking type and the proportion of the covered area and execute an early warning measure corresponding to the classification result.

[0059] In some possible embodiments, constructing the high-steep slope stacking sample library includes: Specifically, the first processing module 201 is used for: Collect sample stacking images of the high-steep slope through a camera device and perform annotation processing on the sample stacking images to obtain a sample library including at least five stacking types.

[0060] In some possible embodiments, classifying the stacking image based on the target scoring function to obtain a category probability includes: Specifically, the second processing module 202 is used for: Extract the feature vector of the stacking image and input the feature vector into the target scoring function to calculate the category probability of at least five stacking types corresponding to the stacking image.

[0061] In some possible embodiments, inputting the category probability and the stacking image into a preset neural network model to determine the stacking type and the proportion of the covered area includes: Specifically, the third processing module 203 is used for: Extract features of the stacking image through the neural network model and match the extracted features with the preset stacking type features to determine the stacking type; Segment the stacking image through the neural network model and calculate the ratio of the stacking area to the entire slope area to obtain the proportion of the covered area.

[0062] In some possible embodiments, a classification result is determined based on the type of piled objects and the proportion of the covered area, and warning measures corresponding to the classification result are executed, including: The fourth processing module 204 is specifically configured to: When it is detected that the proportion of the covered area is greater than or equal to a preset first ratio, calculate the distance to the slope edge; When it is detected that the distance to the slope edge is less than or equal to a preset distance, trigger an orange warning light and send a text message notification to the management personnel.

[0063] In some possible embodiments, a classification result is determined based on the type of piled objects and the proportion of the covered area, and warning measures corresponding to the classification result are further executed, including: The fourth processing module 204 is specifically configured to: When it is detected that the proportion of the covered area is greater than or equal to a preset second ratio, trigger a red warning light and generate an emergency work order.

[0064] In some possible embodiments, after executing the warning measures corresponding to the classification result, it further includes: The fourth processing module 204 is specifically configured to: Reconstruct the piled object image through an autoencoder, and calculate the pixel-level mean square error of at least two regions; If there is a continuous region where the pixel-level mean square error is greater than or equal to a preset threshold, mark the piled object image as invalid data.

[0065] Please refer to Figure 3 , Figure 3 which shows a schematic structural diagram of another intelligent recognition and warning device for piled objects on a high-steep slope provided by an embodiment of the present application.

[0066] As Figure 3 shown, the intelligent recognition and warning device 300 for piled objects on a high-steep slope may include at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0067] Among them, the communication bus 302 can be used to realize the connection and communication of the above-mentioned various components.

[0068] Among them, the user interface 303 may include buttons, and the optional user interface may further include a standard wired interface and a wireless interface.

[0069] Among them, the network interface 304 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0070] Among them, the processor 301 may include one or more processing cores. The processor 301 connects various parts within the intelligent recognition and early warning device 300 for high-steep slope piled objects through various interfaces and lines. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling the data stored in the memory 305, it executes various functions of the intelligent recognition and early warning device 300 for high-steep slope piled objects and processes data. Optionally, the processor 301 may be implemented in at least one of the hardware forms of DSP, FPGA, and PLA. The processor 301 may integrate one or a combination of several of CPU, GPU, and modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor 301 and may be implemented separately through a single chip.

[0071] Among them, the memory 305 may include RAM and may also include ROM. Optionally, the memory 305 includes a non-transitory computer-readable medium. The memory 305 can be used to store instructions, programs, codes, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store the data involved in the above-mentioned various method embodiments. Optionally, the memory 305 may also be at least one storage device located far from the aforementioned processor 301. As Figure 3 shown, the memory 305, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an intelligent recognition and early warning application program for high-steep slope piled objects.

[0072] Specifically, the processor 301 may be used to call the intelligent recognition and early warning application program stored in the memory 305 and specifically perform the following operations: Construct a high-steep slope piled object sample library, and based on the high-steep slope piled object sample library, adjust the first parameter of the initial scoring function to obtain a target scoring function including a second parameter; Obtain a piled object image, and classify the piled object image based on the target scoring function to obtain a class probability; Input the class probability and the piled object image into a preset neural network model to determine the piled object type and the proportion of the covered area; Determine a classification result based on the piled object type and the proportion of the covered area, and execute a warning measure corresponding to the classification result.

[0073] The present application also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the above method are implemented. Among them, the computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0074] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0075] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0076] In the several embodiments provided by the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some service interfaces. The indirect couplings or communication connections of the device or unit can be in an electrical or other form.

[0077] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0078] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0079] When an integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. The aforementioned memory includes: USB flash drives, read-only memory (ROM), random access memory (RAM), external hard drives, magnetic disks, or optical discs, etc., which are various media that can store program codes.

[0080] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory, and the memory can include: flash drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs, etc.

[0081] The above are only exemplary embodiments of the present disclosure and should not be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. After considering the specification and practicing the present disclosure, those skilled in the art will easily think of other embodiments of the present disclosure. The present application aims to cover any variations, uses, or adaptive changes of the present disclosure. These variations, uses, or adaptive changes follow the general principles of the present disclosure and include common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. An intelligent identification and early warning method for high and steep slope piles, characterized in that: include: Constructing a sample library of high and steep slope piles, and adjusting a first parameter of an initial scoring function based on the sample library of high and steep slope piles to obtain a target scoring function including a second parameter; Acquire a pile image, and classify the pile image based on the target score function to obtain a category probability; Inputting the category probability and the pile image into a preset neural network model to determine the pile type and coverage area ratio; A classification result is determined based on the type of the pile and the coverage area ratio, and early warning measures corresponding to the classification result are executed.

2. The method according to claim 1, characterized in that The construction of a high and steep slope pile sample library comprises: Sample pile images of high and steep slopes are collected by camera equipment, and the sample pile images are annotated to obtain a sample library containing at least five types of piles.

3. The method according to claim 1, characterized in that The classifying the pile image based on the target score function to obtain the category probability includes: A feature vector of the object pile image is extracted, and the feature vector is input into the target score function to calculate and obtain category probabilities of at least five types of objects piled corresponding to the object pile image.

4. The method according to claim 1, characterized in that The step of inputting the category probability and the pile image into a preset neural network model to determine the pile type and the coverage area ratio includes: Extracting features of the pile image by using the neural network model, and matching the extracted features with preset pile type features to determine the pile type; The pile image is segmented by the neural network model, the ratio of the pile area to the entire slope area is calculated, and the coverage area ratio is obtained.

5. The method according to claim 1, characterized in that The determining of the classification result based on the type of the pile and the coverage area ratio, and executing the early warning measures corresponding to the classification result, includes: When it is detected that the coverage area ratio is greater than or equal to a preset first ratio, calculating the distance to the edge of the slope; When it is detected that the distance to the edge of the slope is less than or equal to the preset distance, an orange warning light is triggered and a text message notification is sent to the management personnel.

6. The method according to claim 1, characterized in that The method of determining a classification result based on the type of the pile and the coverage area ratio, and executing early warning measures corresponding to the classification result, further includes: When it is detected that the coverage area ratio is greater than or equal to a preset second ratio, a red warning light is triggered and an emergency work order is generated.

7. The method according to claim 1, characterized in that After executing the early warning measures corresponding to the classification results, the method further includes: Reconstructing the pile image by an autoencoder, and calculating pixel-level mean square errors of at least two regions; If there is a continuous area where the pixel-level mean square error is greater than or equal to a preset threshold, the pile image is marked as invalid data.

8. An intelligent identification and early warning device for piles on high and steep slopes, characterized in that: include: A first processing module is used to construct a sample library of high and steep slope piles, and adjust a first parameter of an initial scoring function based on the sample library of high and steep slope piles to obtain a target scoring function including a second parameter; A second processing module is used to obtain a pile image, and classify the pile image based on the target score function to obtain a category probability; A third processing module is used to input the category probability and the pile image into a preset neural network model to determine the pile type and coverage area ratio; The fourth processing module is used to determine a classification result based on the type of the pile and the coverage area ratio, and execute early warning measures corresponding to the classification result.

9. An intelligent identification and early warning device for piles on high and steep slopes, characterized in that: including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a computer or a processor, the computer or the processor executes the steps of the method according to any one of claims 1 to 7.

Images