Distributed construction waste intelligent identification method and system based on adversarial network

Abstract

The invention discloses a distributed construction waste intelligent identification method and system based on an adversarial network, and relates to the field of image identification. The method comprises the steps that a target area is acquired, and a construction waste knowledge base is built; overlooking shooting is carried out, an overlooking area image is obtained, the overlooking area image is cut, and a plurality of image blocks are obtained; arranging sensors in the plurality of image blocks to obtain image block information; an exclusive image recognition model is built for the image blocks, a construction waste intelligent recognition module is formed based on a distributed framework, and the construction waste intelligent recognition module is in communication connection with the construction waste knowledge base; and acquiring real-time construction waste image information, inputting the real-time construction waste image information into the intelligent construction waste identification module for identification, and outputting an identification result. By adopting the method, the technical problems that in the prior art, the construction waste identification technology is poor, and the condition of the construction waste is difficult to accurately identify are solved, and the technical effect of efficiently and accurately identifying the construction waste is achieved.

Description

Technical Field

[0001] This application relates to the field of image recognition technology, especially the field of construction waste image recognition. Specifically, it is a distributed intelligent construction waste recognition method and system based on adversarial networks. Background Art

[0002] An adversarial network, also known as a generative adversarial network, is a deep learning model used to generate a large amount of construction waste image data, thereby expanding the training dataset and improving the generalization ability of the recognition model. A distributed system can store construction waste image data dispersedly on multiple nodes, and each node runs one or more recognition models. In the prior art, the recognition of construction waste is slow and the accuracy is poor, resulting in untimely waste treatment and causing adverse environmental impacts.

[0003] In summary, there is a technical problem in the prior art that the construction waste recognition technology is poor and it is difficult to accurately recognize the situation of construction waste. Summary of the Invention

[0004] Based on this, it is necessary to provide a distributed intelligent construction waste recognition method and system based on adversarial networks that can achieve efficient and accurate recognition of construction waste for the above technical problems.

[0005] In the first aspect, a distributed intelligent construction waste recognition method based on adversarial networks is provided. The method includes: obtaining a target area, collecting historical data based on the target area, and building a construction waste knowledge base; taking an overhead shot of the target area to obtain an overhead area image, cutting the overhead area image to obtain multiple image blocks; arranging sensors in the multiple image blocks for multi-angle collection to obtain image block information; building an exclusive image recognition model for the image blocks. Based on a distributed framework, the exclusive image recognition models are interconnected to form a construction waste intelligent recognition module, and the construction waste intelligent recognition module is communicatively connected to the construction waste knowledge base; obtaining real-time construction waste image information based on the sensors, inputting it into the construction waste intelligent recognition module for recognition, and outputting a recognition result.

[0006] Second aspect, a distributed intelligent recognition system for construction waste based on an adversarial network, the system comprising: a construction waste knowledge base building module for obtaining a target area, collecting historical data based on the target area, and building a construction waste knowledge base; an overhead area image cutting module for taking an overhead shot of the target area, obtaining an overhead area image, and cutting the overhead area image to obtain a plurality of image blocks; an image block information obtaining module for arranging sensors in the plurality of image blocks, collecting from multiple angles, and obtaining image block information; a construction waste intelligent recognition module building module for building an exclusive image recognition model for the image blocks, and based on a distributed framework, the exclusive image recognition models are connected to each other to form a construction waste intelligent recognition module, and the construction waste intelligent recognition module is communicatively connected to the construction waste knowledge base; and a recognition result output module for obtaining real-time construction waste image information based on the sensors, inputting the real-time construction waste image information into the construction waste intelligent recognition module for recognition and outputting a recognition result.

[0007] Third aspect, there is provided a computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps described in the first aspect when executing the computer program.

[0008] Fourth aspect, there is provided a computer-readable storage medium having stored thereon a computer program, and the computer program, when executed by a processor, implements the steps described in the first aspect.

[0009] The above-mentioned distributed intelligent recognition method and system for construction waste based on an adversarial network solve the technical problem in the prior art of poor construction waste recognition technology and difficulty in accurately recognizing construction waste, and achieve the technical effect of efficiently and accurately recognizing construction waste.

[0010] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented in accordance with the content of the specification. And in order to make the above and other objects, features and advantages of the present application more obvious and understandable, the following specifically gives the specific embodiments of the present application. Description of the Drawings

[0011] Figure 1 It is a schematic flow chart of a distributed intelligent recognition method for construction waste based on an adversarial network in an embodiment; Figure 2 It is a schematic flow chart of obtaining error data of a distributed intelligent recognition method for construction waste based on an adversarial network in an embodiment; Figure 3It is a structural block diagram of a distributed intelligent recognition system for construction waste based on an adversarial network in an embodiment; Figure 4 It is an internal structure diagram of a computer device in an embodiment.

[0012] Explanation of reference numerals: Construction waste knowledge base building module 11, top-down area image cutting module 12, image block information acquisition module 13, construction waste intelligent recognition module construction module 14, recognition result output module 15. Specific implementation manners

[0013] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] As Figure 1 shown, the present application provides a distributed intelligent recognition method for construction waste based on an adversarial network, and the method includes: Obtain a target area, collect historical data based on the target area, and build a construction waste knowledge base; An adversarial network is a deep learning model, and its feature lies in the adversarial competition between two networks. The competition mechanism enables the two networks to be continuously optimized during the training process, thereby improving the authenticity of the generated data and the accuracy of the discriminator; a distributed system distributes data on multiple nodes for parallel processing, which can significantly improve the computing efficiency and recognition speed. The distributed intelligent recognition technology for construction waste based on an adversarial network combines the advantages of deep learning and distributed systems, and realizes efficient and accurate recognition and classification of construction waste in practical applications, providing strong support for the resource utilization and environmental protection of construction waste.

[0015] The target area refers to the area where construction waste is placed for research. Defining the target area helps subsequent data collection and knowledge base construction work; obtaining the past construction waste information in the area according to the target area, such as the amount of construction waste generated, types, treatment methods, transportation routes, etc., through public data, reports of relevant enterprises, etc., and building a construction waste knowledge base according to the historical data. The construction waste knowledge base includes images of the construction waste and the corresponding annotation information. Collect historical data based on the target area and build a comprehensive, accurate and practical construction waste knowledge base to provide strong support for the recognition and management of construction waste.

[0016] Build an adversarial network in the construction waste knowledge base, and the adversarial network includes a generation model and a discriminant model; Alternately optimize and train according to the adversarial network to obtain an optimized generation model, and obtain similar real construction waste data based on the generation model to expand the construction waste knowledge base.

[0017] Build an adversarial network in the construction waste knowledge base. The adversarial network includes a generation model and a discriminative model. The task of the generation model is to generate fake data that is as close as possible to real construction waste data. A convolutional neural network is embedded in the generation model to generate new data samples based on the distribution of real construction waste data. The task of the discriminative model is to distinguish whether the input construction waste data is real or generated by the generation model. During the training process, the generation model and the discriminative model are alternately optimized and trained. Specifically, the generation model is fixed and the discriminative model is trained. The discriminative model tries to distinguish between real data extracted from the construction waste knowledge base and fake data generated by the generation model. By minimizing the classification error, the discriminative model improves its discrimination ability. The goal of the generation model is to generate data that can "fool" the discriminative model, that is, to make the discriminative model misjudge it as real data. By optimizing the generation model, the data generated by it becomes closer and closer to real data in the view of the discriminative model. After alternately optimizing and training, the generation model is optimized. Using the optimized generation model, a large amount of similar real construction waste data can be generated. Although the similar real construction waste data is generated by the model, it is very close to the real data in terms of structure and features, so it can be regarded as effective expanded data. Adding these generated data to the construction waste knowledge base can further enrich the content of the knowledge base and improve its coverage and integrity. In this application, the application of the adversarial network is mainly reflected in improving the recognition accuracy and robustness. Through adversarial training, the generation model can learn more features of construction waste, so as to more accurately identify different types of waste. At the same time, the introduction of adversarial samples can also help the model improve its robustness to interference factors such as noise and occlusion, making the model more reliable in practical applications. Build and optimize an adversarial network in the construction waste knowledge base, and use the generation model to expand the knowledge base to provide more comprehensive and accurate data support for the identification and management of construction waste.

[0018] Take a top-down photo of the target area to obtain a top-down area image, and cut the top-down area image to obtain multiple image blocks; Use devices such as drones, satellites, or ground high-angle cameras to take a top-down view of the target area, obtaining a top-down area image of the target area. The top-down area image can capture the overall layout, terrain features, and the distribution of construction waste in the target area. According to the cutting rules, cut the preprocessed top-down area image, dividing the entire image into multiple smaller image blocks for subsequent more refined processing and analysis. After the cutting is completed, multiple image blocks will be obtained, and the image blocks need to be numbered, marked, etc. for subsequent management and use. Obtain multiple image blocks containing different parts of the target area, providing basic data support for subsequent construction waste identification work.

[0019] Query the information of the target area, adjust the shooting height based on the information of the target area, and obtain a top-down area image; Perform image preprocessing based on the top-down area image to obtain a processed image; According to the processed image and the information of the target area, perform fixed-area grid division to obtain a boundary area; Perform image segmentation based on the boundary area to obtain multiple image blocks.

[0020] Query the specific information of the target area through the Geographic Information System (GIS), including the area range, topography, building distribution, etc. The target area information will be used as an important basis for adjusting the shooting height and image processing. According to the queried target area information, determine an appropriate shooting height. The selection of the shooting height should consider factors such as the size of the target area, the complexity of the terrain, and the resolution of the required images. By adjusting the shooting height of the drone or satellite, ensure that a clear and complete top-down area image can be obtained. After adjusting the shooting height, conduct a top-down shooting to obtain an image of the target area. Ensure that the image can cover the entire target area and the details are clearly distinguishable. Preprocess the obtained top-down area image to eliminate noise, improve contrast, correct color deviation, etc., to improve the visual quality of the image and the accuracy of subsequent processing; combine the target area information to divide the preprocessed image into fixed-area grids. The area of each grid can be set according to needs to ensure that the amount of construction waste information contained in each grid is moderate, facilitating subsequent processing and analysis. The grid division should consider the shape and size of the target area to ensure that the grid can completely cover the entire target area. Based on the grid division, determine the boundary areas of each grid. These boundary areas will be used as the basis for image segmentation to ensure that each image block contains complete grid information; based on the determined boundary areas, segment the image to obtain multiple image blocks. Each image block corresponds to a grid area and contains the construction waste information within that area. Obtain high-quality top-down area images based on the target area information and conduct effective image processing and segmentation to provide reliable data support for subsequent construction waste identification and classification work.

[0021] Deploy sensors in the multiple image blocks, collect from multiple angles, and obtain image block information; Conduct a detailed analysis of each image block to understand its terrain, construction waste distribution, and characteristics, and then plan the deployment positions and angles of the sensors to ensure that the information within the image block can be comprehensively collected. Start the sensors and collect from multiple angles. By adjusting the angles and heights of the sensors, obtain image block information from different perspectives. During the collection process, ensure the stability and collection speed of the sensors to obtain continuous and clear image data; obtain the corresponding image block information according to the multiple image blocks.

[0022] Build an exclusive image recognition model for the image blocks. Based on a distributed framework, the exclusive image recognition models are interconnected to form a construction waste intelligent recognition module, and the construction waste intelligent recognition module is communicatively connected to the construction waste knowledge base; First, in-depth feature extraction and analysis are performed on each image patch. This includes extracting features such as color, texture, and shape in the image patch. For each image patch, a dedicated image recognition model is built. The dedicated image recognition model uses a convolutional neural network and is optimized according to the characteristics of the features of the image patch and the types of construction waste to ensure the accuracy and efficiency of the dedicated image recognition model. A distributed framework is used to connect the dedicated image recognition models to ensure that the dedicated image recognition models can process data in parallel and share information and resources in real time. The models are integrated under the distributed framework to form a construction waste intelligent recognition module. The construction waste intelligent recognition module can receive the input image patch data, automatically call the corresponding dedicated image recognition model for recognition, and output the recognition result. The construction waste intelligent recognition module needs to establish a communication connection with the construction waste knowledge base. The construction waste intelligent recognition module can access and update the information in the knowledge base in real time. The construction waste intelligent recognition module obtains relevant information from the construction waste knowledge base to assist the recognition process. A construction waste intelligent recognition module based on a distributed framework is built and a communication connection with the construction waste knowledge base is realized, providing strong technical support for the recognition and classification of construction waste.

[0023] Extract the historical data of the image patches based on the construction waste knowledge base to obtain a historical data set; Segment the historical data set to obtain a training set, a validation set, and a test set; Build an initial dedicated image recognition model based on a convolutional neural network and perform supervised training according to the training set, the validation set, and the test set; Preset an accuracy index. When the model accuracy of the initial dedicated image recognition model meets the preset accuracy index, obtain the dedicated image recognition model.

[0024] Utilize the historical data resources in the construction waste knowledge base to extract the historical data of image blocks, including the recognition results, annotation information, feature extraction results, etc. of construction waste images in the past time period, obtain the historical data set, and segment the extracted historical data set, usually divided into a training set, a validation set, and a test set. The training set is used to train the model, the validation set is used to adjust the model parameters and select the best model during the training process, and the test set is used to evaluate the performance of the model. Specifically, generally, it is segmented according to the ratio of 70% training set, 15% validation set, and 15% test set. The specific ratio can be adjusted according to the data volume and task requirements; construct an initial exclusive image recognition model based on the convolutional neural network. The convolutional neural network has excellent performance and can effectively extract image features and classify them; use the segmented training set to perform supervised training on the initial exclusive image recognition model. During the training process, optimize the model parameters through the backpropagation algorithm so that the initial exclusive image recognition model can accurately recognize construction waste images; preset an accuracy rate index as the goal of model training, and the accuracy rate index should be set according to actual needs. For example, it can be set to an accuracy rate of 90%; when the model accuracy reaches or exceeds the preset accuracy rate index, it is considered that the model training is completed, and an exclusive image recognition model is obtained. The exclusive image recognition model has a high recognition accuracy. Construct an exclusive recognition model for construction waste image blocks based on the construction waste knowledge base, providing effective technical support for subsequent construction waste recognition and classification tasks.

[0025] Select a distributed framework according to the information of the target area; Deploy the exclusive image recognition model into the distributed framework to ensure that the exclusive image recognition model can run independently; Identify the connection area, perform error identification according to the connection area, and obtain error data; Perform fitting based on the error data to obtain a construction waste intelligent recognition module.

[0026] Based on the characteristics and requirements of the target area, a distributed framework is selected. For example, when the target area is wide in scope and has a huge amount of data, a distributed framework that can support large-scale data processing is selected. After the distributed framework is selected, the previously constructed exclusive image recognition model is deployed into the distributed framework according to the segmented images. For example, if the target area is equally divided into four grids, then the number of nodes in the distributed framework is also four. The exclusive image recognition model is imported into the distributed framework according to the corresponding image blocks to obtain the intelligent construction waste recognition module after completion. Identify the connection area, perform error recognition based on the connection area to obtain error data, where the connection area refers to the area where multiple image blocks intersect. The errors include construction waste segmentation, image quality, and occlusion conditions. Since the target area in this application is divided into grids of a fixed area, it is possible that a complete construction waste segmentation is in adjacent image blocks, denoted as construction waste segmentation. Image quality refers to the problem of unclear splicing caused by abnormal situations during image segmentation. The occlusion condition refers to the situation where construction waste is occluded due to sunshades and other conditions in the open-air environment, resulting in image errors. Based on the error data, fitting can be performed. These data can be used to fit the intelligent construction waste recognition module. The purpose of fitting is to adjust the parameters and structure of the model to reduce errors and improve recognition accuracy. In this application, it means making a judgment based on the error data and proposing error compensation to improve the construction waste recognition accuracy, obtaining the intelligent construction waste recognition module. The intelligent construction waste recognition module can perform efficient parallel processing based on the distributed framework and can accurately identify construction waste images. Select a suitable distributed framework according to the target area information and deploy the image recognition model to construct and optimize the intelligent construction waste recognition module.

[0027] As Figure 2 shown, obtain the boundary area, perform information extraction based on the boundary area to obtain the extracted information; Perform error recognition based on the extracted information. The errors include construction waste segmentation, image quality, and occlusion conditions; Perform data integration based on the errors to obtain error data.

[0028] Obtain the boundary region, which refers to the region where the multiple image patches intersect and contains various information. After obtaining the boundary region, further extract information from the boundary region, including features such as the shape, texture, and color of the boundary, as well as the relative position relationship with the surrounding environment, etc. After extracting the information of the boundary region, it is necessary to perform error identification on the information, including construction waste segmentation, image quality, and the situation of occlusions, judge the attribution type of the error, perform data integration, obtain the boundary error situation of the construction waste intelligent recognition module, and integrate the error situation to obtain error data. Based on the boundary region, information extraction and error identification are carried out to obtain error data, which provides an important basis for subsequent model optimization and improvement of recognition performance.

[0029] Judge the error data to obtain the types of error data. Based on the types of error data, perform the connection region compensation to obtain the compensation result. According to the compensation result, perform fitting to obtain the construction waste intelligent recognition module.

[0030] Carefully analyze the error data obtained, judge the type of error data it belongs to, based on the type of error data, perform the connection region compensation to obtain the compensation result. The purpose of compensation is to correct errors and improve the accuracy of recognition. If the error data belongs to construction waste segmentation, then combine the adjacent image patches, and build a dedicated image recognition model for the combined image patches based on the above method. If the error type is image quality, then use image enhancement technology to improve the image quality. If the error type is the situation of occlusions, then adjust the shooting angle, take images again and analyze, judge the type of error data until the judgment is completed and the compensation result is generated. According to the compensation result, perform fitting, and the fitting process includes adjusting the parameters of the model, updating the structure of the model, etc., to obtain the construction waste intelligent recognition module. Through fitting, the model can better adapt to the error data and improve the overall recognition performance. After completing the fitting, the optimized construction waste intelligent recognition module can be obtained. Perform connection region compensation according to the type of error data, and fit and optimize the construction waste intelligent recognition module based on the compensation result, so as to improve the accuracy and efficiency of recognition.

[0031] Based on the sensor, obtain real-time construction waste image information, input it into the construction waste intelligent recognition module for recognition and output the recognition result.

[0032] Based on the sensors, real-time construction waste image information is obtained. According to the sensors in the target area, such as cameras, etc., the image information of construction waste is captured in real time, and the real-time construction waste image information is input into the intelligent construction waste recognition module; after receiving the image information, the intelligent construction waste recognition module will automatically allocate it to the corresponding image blocks for exclusive image recognition, including extracting the features in the image, matching them with the knowledge learned in the model, and outputting the final recognition result. By using the real-time construction waste image information obtained by the sensors and performing real-time recognition and output through the intelligent construction waste recognition module, it provides effective technical support for the management and resource utilization of construction waste.

[0033] Such as Figure 3 As shown in the figure, the embodiment of the present application includes a distributed intelligent construction waste recognition system based on an adversarial network. The system includes: A construction waste knowledge base building module 11, which is used to obtain the target area, collect historical data based on the target area, and build a construction waste knowledge base; An overhead area image cutting module 12, which is used to take an overhead view of the target area, obtain an overhead area image, and cut the overhead area image to obtain multiple image blocks; An image block information acquisition module 13, which is used to deploy sensors in the multiple image blocks, collect from multiple angles, and obtain image block information; A construction waste intelligent recognition module construction module 14, which is used to build an exclusive image recognition model for the image blocks. Based on a distributed framework, the exclusive image recognition models are interconnected to form a construction waste intelligent recognition module. The construction waste intelligent recognition module is communicatively connected to the construction waste knowledge base; A recognition result output module 15, which is used to obtain real-time construction waste image information based on the sensors, input it into the construction waste intelligent recognition module for recognition and output the recognition result.

[0034] Furthermore, the embodiment of the present application further includes: An adversarial network construction module, which is used to build an adversarial network in the construction waste knowledge base. The adversarial network includes a generative model and a discriminative model; A construction waste knowledge base expansion module, which is used to perform alternating optimization training according to the adversarial network, obtain an optimized generative model, and obtain similar real construction waste data based on the generative model to expand the construction waste knowledge base.

[0035] Further, the embodiments of the present application further include: An overhead area image acquisition module, which is used to query information about the target area, adjust the shooting height based on the information about the target area, and acquire an overhead area image; A processed image acquisition module, which is used to perform image preprocessing on the overhead area image to acquire a processed image; A boundary area acquisition module, which is used to perform fixed-area grid division according to the processed image in combination with the information about the target area to acquire a boundary area; A plurality of image block acquisition modules, which are used to perform image segmentation based on the boundary area to acquire a plurality of image blocks.

[0036] Further, the embodiments of the present application further include: A historical data set acquisition module, which is used to extract historical data of image blocks based on the construction waste knowledge base to acquire a historical data set; A historical data set segmentation module, which is used to segment the historical data set to acquire a training set, a validation set, and a test set; A supervised training module, which is used to construct an initial exclusive image recognition model based on a convolutional neural network and perform supervised training according to the training set, the validation set, and the test set; An exclusive image recognition model acquisition module, which is used to preset an accuracy rate index, and when the model accuracy of the initial exclusive image recognition model meets the preset accuracy rate index, acquire the exclusive image recognition model.

[0037] Further, the embodiments of the present application further include: A distributed framework selection module, which is used to select a distributed framework according to the information about the target area; An independent operation module, which is used to deploy the exclusive image recognition model into the distributed framework to ensure that the exclusive image recognition model can operate independently; An error data acquisition module, which is used to identify a connection area, perform error identification according to the connection area, and acquire error data; An error data fitting module, which is used to perform fitting based on the error data to acquire a construction waste intelligent recognition module.

[0038] Further, the embodiments of the present application further include: An extraction information acquisition module, which is used to acquire a boundary region, extract information based on the boundary region, and obtain extraction information; An error identification module, which is used to identify errors based on the extraction information, and the errors include construction waste segmentation, image quality, and occluder conditions; An error data integration module, which is used to integrate data based on the errors to obtain error data.

[0039] Furthermore, the embodiments of the present application further include: An error data type acquisition module, which is used to judge the error data and obtain the error data type; A connection area compensation module, which is used to perform the connection area compensation based on the error data type to obtain a compensation result; A compensation result fitting module, which is used to perform fitting according to the compensation result to obtain the construction waste intelligent identification module.

[0040] For the specific embodiments of the distributed construction waste intelligent identification system based on the adversarial network, reference can be made to the embodiments of the distributed construction waste intelligent identification method based on the adversarial network in the above text, which will not be elaborated here. The above modules can be embedded in the processor in the computer device in the form of hardware or independent of it, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0041] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 4 shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store news data and data such as time decay factors. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the distributed construction waste intelligent identification method based on the adversarial network.

[0042] Those skilled in the art can understand, Figure 4The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0043] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps of the distributed construction waste intelligent recognition method based on the adversarial network are implemented.

[0044] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the steps of the distributed construction waste intelligent recognition method based on the adversarial network are implemented.

[0045] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0046] The above-described embodiments only represent several implementation manners of this application, and their descriptions are relatively specific and detailed, but they should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application patent shall be subject to the appended claims.

Claims

1. A distributed intelligent identification method for construction waste based on adversarial networks, characterized in that: The method comprises: Acquire a target area, collect historical data based on the target area, and build a construction waste knowledge base; Taking a bird's-eye view of the target area to obtain a bird's-eye view area image, and cutting the bird's-eye view area image to obtain a plurality of image blocks; Arrange sensors in the multiple image blocks, perform multi-angle acquisition, and obtain image block information; A dedicated image recognition model is constructed for the image block. Based on a distributed framework, the dedicated image recognition models are interconnected to form a construction waste intelligent recognition module. The construction waste intelligent recognition module is communicatively connected with the construction waste knowledge base. Real-time construction waste image information is acquired based on the sensor, input into the construction waste intelligent recognition module for recognition and output of the recognition result.

2. The method according to claim 1, characterized in that Methods include: Building an adversarial network in the construction waste knowledge base, wherein the adversarial network includes a generative model and a discriminative model; Alternating optimization training is performed according to the adversarial network to obtain an optimized generation model, similar real construction waste data is obtained based on the generation model, and the construction waste knowledge base is expanded.

3. The method according to claim 1, characterized in that The target area is photographed from above to obtain an image of the overhead area, and the image of the overhead area is cut to obtain a plurality of image blocks, the method comprising: Querying the information of the target area, adjusting the shooting height based on the information of the target area, and acquiring an image of the overlooking area; Performing image preprocessing based on the overhead area image to obtain a processed image; Performing fixed-area grid division according to the processed image combined with information of the target area to obtain a boundary area; Image segmentation is performed based on the boundary area to obtain multiple image blocks.

4. The method according to claim 1, characterized in that A dedicated image recognition model is constructed for the image block. Based on a distributed framework, the dedicated image recognition models are interconnected to form a construction waste intelligent recognition module. The construction waste intelligent recognition module is communicatively connected with the construction waste knowledge base. The method includes: Extracting historical data of image blocks based on the construction waste knowledge base to obtain a historical data set; Segmenting the historical data set to obtain a training set, a validation set, and a test set; Constructing an initial exclusive image recognition model based on a convolutional neural network, and performing supervised training based on the training set, the validation set, and the test set; A preset accuracy index is used, and when the model accuracy of the initial exclusive image recognition model meets the preset accuracy index, the exclusive image recognition model is obtained.

5. The method according to claim 4, characterized in that include: Selecting a distributed framework according to information of the target area; Deploy the dedicated image recognition model to a distributed framework to ensure that the dedicated image recognition model can run independently; Identify the connection area, perform error identification based on the connection area, and obtain error data; Fitting is performed based on the error data to obtain a construction waste intelligent identification module.

6. The method according to claim 5, characterized in that Identify the connection area, perform error identification according to the connection area, and obtain error data, the method comprising: Acquire a boundary area, extract information based on the boundary area, and acquire extracted information; Perform error identification based on the extracted information, wherein the errors include construction waste segmentation, image quality, and occlusion conditions; Data integration is performed based on the error to obtain error data.

7. The method according to claim 6, characterized in that Performing data integration based on the error to obtain error data, the method includes: Judging the error data to obtain the type of error data; Performing connection area compensation based on the error data type to obtain a compensation result; Fitting is performed according to the compensation result to obtain the construction waste intelligent identification module.

8. A distributed intelligent identification system for construction waste based on adversarial networks, characterized in that: The system comprises: A construction waste knowledge base building module, the construction waste knowledge base building module is used to obtain a target area, collect historical data based on the target area, and build a construction waste knowledge base; A bird's-eye view area image cutting module, the bird's-eye view area image cutting module is used to take a bird's-eye view photo of the target area, obtain a bird's-eye view area image, and cut the bird's-eye view area image to obtain a plurality of image blocks; An image block information acquisition module, the image block information acquisition module is used to deploy sensors in the multiple image blocks, perform multi-angle acquisition, and acquire image block information; A construction waste intelligent identification module construction module, the construction waste intelligent identification module construction module is used to build an exclusive image recognition model for the image block, based on a distributed framework, the exclusive image recognition models are interconnected to form a construction waste intelligent identification module, and the construction waste intelligent identification module is communicatively connected with the construction waste knowledge base; The recognition result output module is used to obtain real-time construction waste image information based on the sensor, input it into the construction waste intelligent recognition module for recognition and output the recognition result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images