Construction method, system and equipment of municipal drainage facility defect database and medium

By constructing a municipal drainage facility database, using multi-label classification confusion matrix and defect analysis model to optimize defect classification, the problem of low defect recognition accuracy in ground drainage facilities in the existing technology is solved, and the database distinction accuracy and data quality are improved.

CN119025693AInactive Publication Date: 2024-11-26GUANGDONG AGCO ENVIRONMENTAL TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411122882.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing municipal drainage facilities defect classification method is difficult to correctly identify the defect types of ground drainage facilities. Due to the problems of intra-class differences and inter-class similarity, the model needs more parameters and levels to extract higher-level features, increasing the calculation amount. and training time.

Method used

By obtaining ground drainage facilities sample data, the initial sample data set is constructed, and using multi-label classification confusion matrix and defect analysis model, the defect classification is optimized, labeling detection and correction is carried out, and a high-quality municipal drainage facility database is formed to reduce in-class differences and Similar issues between classes.

Benefits of technology

The municipal drainage facility database has improved the accuracy of distinguishing the defect types of facilities, improved the mixed quenching of the defect types of database, and improved the data quality and labeling detection accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025693A_ABST
    Figure CN119025693A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of data processing, in particular to a construction method, system and equipment for a database lacking of municipal drainage facilities and a medium, and the construction method for the database lacking of the municipal drainage facilities comprises the following steps: obtaining ground drainage facility sample data, and constructing the ground drainage facility sample data into an initial sample data set; obtaining a real defect type set and a predicted defect type set according to the initial sample data set, constructing a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, and inputting the initial sample data set into the multi-label classification confusion matrix to obtain defect analysis data; obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification; marking the ground drainage facility data after defect classification to obtain marked ground drainage facility data, inputting the marked ground drainage facility data into a preset defect analysis model, and repeatedly detecting the marked ground drainage facility data to obtain a defect marking detection result, correcting the labeled ground drainage facility data according to the labeled detection result; and integrating the corrected ground drainage facility data with the marking box to form a municipal drainage facility database. The method has the effects of improving the distinguishing precision of the municipal drainage facility database for the facility defect types and improving the confusion condition of the database defect types.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of drainage data processing, and in particular to a method, system, equipment and medium for constructing a municipal drainage facility database. Background Art

[0002] With the rapid development of computer vision technology, domestic and foreign researchers have used computer vision technology to do a lot of work on underground drainage pipe defect detection, but there is still a blank in the research on urban ground drainage facility defect detection. Therefore, it is urgent to use computer vision technology to intelligently identify the target state of ground drainage facility defects.

[0003] In the process of building a municipal drainage facility database, before using samples to train the model and obtain the database, the collected photos of municipal drainage facilities are first classified and processed. When manually labeling the samples, the staff can mark the defect types on the samples according to the pre-classification. In this way, when the machine learns the samples through image recognition, it is easier for the machine to summarize and conclude the defects of ground drainage facilities, which helps to improve the accuracy of the database.

[0004] However, according to the above industry standards and specifications such as "Technical Specifications for Data Collection and Maintenance of Urban Drainage and Flood Control Facilities", "Technical Regulations for Inspection and Evaluation of Urban Drainage Pipes", and "Technical Regulations for Operation, Maintenance and Safety of Urban Drainage Pipes and Pumping Stations", there are problems of intra-class differences and inter-class similarities in the classification of defects in ground drainage facilities. For example, the defect characteristics of blockage inside the rain grate and blockage of the rain grate holes are similar, both of which manifest as objects blocking the grate holes; the defect characteristics of damaged rain grate frames and damaged rain grate are also relatively similar, both of which manifest as damage to the surface morphology of the rain grate; there is a high possibility of type confusion between the above defects. Therefore, when using the existing ground drainage facility classification method to classify samples, it is difficult for the machine to correctly identify and learn the defect types of ground drainage facilities, and in order to cope with intra-class differences and inter-class similarities, the model requires more parameters and levels to extract higher-level features, which will increase the computational complexity and training time of the model. Therefore, there is room for improvement. Summary of the invention

[0005] In order to improve the accuracy of distinguishing facility defect types in a municipal drainage facility database and improve the confusion of database defect types, the present application provides a method, system, equipment and medium for constructing a municipal drainage facility defect database.

[0006] The above-mentioned invention objective of the present application is achieved through the following technical solutions: A method for constructing a municipal drainage facility deficiency database, the method comprising the steps of: acquiring surface drainage facility sample data, and constructing the surface drainage facility sample data into an initial sample data set; Acquire a real defect type set and a predicted defect type set according to the initial sample data set, input the real defect type set and the predicted defect type set into a preset multi-label classification confusion matrix, and obtain defect analysis data; Obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification; Labeling the defect-classified ground drainage facility data to obtain labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model to obtain defect labeling detection results, and correcting the labeled ground drainage facility data according to the labeling detection results; The corrected surface drainage facility data with marked boxes are integrated to form a municipal drainage facility database.

[0007] By adopting the above technical solution, in the process of building a municipal drainage facility database, a large amount of surface drainage facility sample data can be obtained through on-site actual working condition images or on the Internet. The surface drainage facility data refers to the surface drainage facility image. The collected surface drainage facility data are integrated to form an initial sample data set. The initial sample data set is defect-classified to obtain a real defect type set and a predicted defect type set. The real defect type set refers to a set of facility defect types obtained by defect-classifying surface drainage facility sample data according to traditional industry standards. The predicted defect type set refers to a set of defect type prediction analysis of surface drainage facility sample data. The real defect type set and the predicted defect type set are input into a preset multi-label classification confusion matrix to obtain defect analysis data. The defect analysis data refers to the multi-label classification confusion matrix, which determines the surface drainage facility samples according to the proportion of the number of correctly predicted samples to the total number of samples. The accuracy of defect classification of data is improved, and the defect classification is optimized using the obtained defect analysis data. According to the optimized defect classification, the sample data of ground drainage facilities are defect-classified to obtain the ground drainage facilities data after defect classification, and then the ground drainage facilities data after defect classification are labeled, and the defect analysis model is used to repeatedly perform labeling detection on the labeled ground drainage facilities data to obtain the labeling detection results. The labeling of the ground drainage facilities data is corrected according to the labeling detection results, and the corrected ground drainage facilities data with labeling boxes are integrated to form a municipal drainage facilities database. During the entire database construction process, the defect classification of the ground drainage facilities data is optimized using a multi-label classification confusion matrix, and the defect labeling of the ground drainage facilities data is detected and corrected using a defect analysis model, which improves the municipal drainage facilities database's accuracy in distinguishing facility defect types and improves the confusion of database defect types.

[0008] In a preferred example, the present application may be further configured as follows: the acquiring of the surface drainage facility sample data and constructing the surface drainage facility sample data into an initial sample data set specifically includes: The actual working condition image of the site is obtained, and the initial ground drainage facility image is crawled on the Internet according to the actual working condition image of the site; the actual working condition image of the site and the initial ground drainage facility image are input into a preset image screening model to obtain initial sample data.

[0009] By adopting the above technical solution, in the process of collecting ground drainage facility data, the staff takes photos on site to obtain the actual working condition images of the site first, and then crawls through the search engine on the Internet to obtain the initial ground drainage facility images based on the obtained actual working condition images of the site. The collected initial ground drainage facility images and the actual working condition images of the site will contain other debris that affect the classification of facility defects. The preset image screening model is used to filter and denoise the actual working condition images of the site and the initial ground drainage facility images to obtain a high-quality initial sample data set to improve the data quality in the database.

[0010] In a preferred example, the present application can be further configured as follows: obtaining a real defect type set and a predicted defect type set according to the initial sample data set, inputting the real defect type set and the predicted defect type set into a preset multi-label classification confusion matrix, and obtaining defect analysis data, specifically including: Based on the preset facility defect classification rules, the initial sample data set is defect-classified to obtain a real defect type set; the initial sample data set is input into the preset defect analysis model to obtain a predicted defect type set; Inputting the true defect type set and the predicted defect type set into the multi-label classification confusion matrix, with the true defect type set as the horizontal column and the predicted type set as the vertical column; The initial sample data is input into a multi-label classification confusion matrix to obtain defect analysis data.

[0011] By adopting the above technical scheme, in the process of optimizing the defect classification of ground drainage facilities, the detection classification is determined by using the facility defect classification rules of traditional industry standards, and the sample data in the initial sample data set is first labeled according to the detection classification to obtain a set of real defect types, and then the sample data in the initial sample data set is input into the defect analysis model, and the predicted defect type set is obtained through the defect analysis model, and the real defect type set is input as a horizontal column and the predicted defect type set as a vertical column into a multi-label classification confusion matrix, and then the sample data in the initial sample data set is input into the multi-label classification confusion matrix to obtain each element value in the matrix, and each element value is used as defect analysis data, so as to optimize the defect classification of ground drainage facilities according to the specific situation of the defect analysis data, improve the accuracy of distinguishing facility defect types in the municipal drainage facility database, and improve the confusion of database defect types.

[0012] In a preferred example, the present application may be further configured as follows: the step of obtaining optimized defect classification based on the defect analysis data and obtaining defect classified ground drainage facility data according to the optimized defect classification specifically includes: Acquire easily confused defect classes according to the defect analysis data, and optimize the multi-label classification confusion matrix based on the easily confused defect classes to obtain an optimized multi-label classification confusion matrix; The defect-classified ground drainage facility data is obtained based on the optimized multi-label classification confusion matrix.

[0013] By adopting the above technical scheme, after the sample data in the initial sample data set is input into the multi-label classification confusion matrix, the value of each element in the multi-label classification confusion matrix is ​​extracted to obtain defect analysis data, and the defect types that are easily confused in the defect classification are determined through the defect analysis data. By analyzing the easily confused defect types and considering the actual application scenarios of operation and maintenance operations, the defect types in the horizontal and vertical columns in the multi-label classification confusion matrix are optimized according to the easily confused defect classes to obtain the optimized multi-label classification confusion matrix, and then the sample data is input into the optimized multi-label classification confusion matrix to obtain the ground drainage facility data with good defect classification, which can effectively reduce the intra-class differences and inter-class similarity problems of the data, and provide important support for building a high-quality ground drainage facility database.

[0014] In a preferred example, the present application may be further configured as follows: the defect-classified ground drainage facility data is labeled to obtain the labeled ground drainage facility data, specifically including: Acquire defect type information based on the defect-classified ground drainage facility data, and acquire defect morphological characteristics and defect range according to the defect type information; A labeling frame is generated based on the defect morphological characteristics and defect range, and the ground drainage facility data after defect classification is labeled to obtain the labeled ground drainage facility data.

[0015] By adopting the above technical scheme, after completing the defect classification of the ground drainage facility data, the corresponding defect type information is obtained according to the classified ground drainage facility data, the defect morphological characteristics and defect range of the ground drainage facility are obtained in the defect type information, and a labeling box is generated according to the defect morphological characteristics and defect range to label the defects of the ground drainage facility data, and the labeled ground drainage facility data is obtained to highlight the defect conditions of the ground drainage facility.

[0016] In a preferred example, the present application can be further configured as follows: the labeled ground drainage facility data is input into a preset defect analysis model, a defect labeling detection result is obtained, and the labeled ground drainage facility data is corrected according to the labeling detection result, specifically including: Input the annotated surface drainage facility data into a preset defect analysis model to obtain defect annotation detection results, wherein the annotation detection results include abnormal annotations and normal annotations; When the marking detection result is a normal marking, retaining the marked ground drainage facility data; When the labeling detection result is an abnormal labeling, labeling abnormality information is obtained based on the labeled ground drainage facility data, the labeling box is corrected according to the labeling abnormality information, and the ground drainage facility data after the defect classification is re-labeled to obtain the labeled ground drainage facility data.

[0017] By adopting the above technical scheme, the labeled ground drainage facility data is labeled and detected through the defect analysis model to obtain the labeling detection result. When the labeling detection result is normal labeling, it means that the defect labeling in the ground drainage facility data after defect classification is accurate, and the labeled ground drainage facility data is retained; when the labeling detection result is abnormal labeling, the labeling abnormality is analyzed to obtain labeling abnormality information. The labeling abnormality information refers to specific labeling problems that occur in the ground drainage facility data, such as missing labeling, type labeling error, inaccurate labeling range box, over-labeling, difficulty in distinguishing the defect range, etc. The labeling box is corrected according to the labeling abnormality information, the ground drainage facility data after defect classification is re-labeled, and the inspection is repeated until the labeling result is normal labeling, so as to improve the process of labeling the ground drainage facility data after defect classification, reduce the probability of labeling abnormalities, and improve the quality of the database and the detection accuracy of data labeling.

[0018] In a preferred example, the present application may be further configured as follows: before the modified ground drainage facility data with the marked frame are integrated to form the municipal drainage facility database, the method for constructing the municipal drainage facility database further comprises the steps of: performing data enhancement processing on the modified ground drainage facility data with the marked frame to obtain enhanced ground drainage facility data; The enhanced surface drainage facility data are integrated to form a municipal drainage facility database.

[0019] By adopting the above technical solution, data enhancement is performed on the corrected ground drainage facility data with marked boxes, such as geometric transformation operations such as image rotation, scaling, cropping, flipping, random erasing, and optical transformation operations such as brightness, contrast, and saturation. It is possible to increase data diversity and improve the quality of the municipal drainage facility database without changing the original data.

[0020] The second object of the invention is achieved by the following technical solutions: A system for constructing a municipal drainage facility missing database, the system comprising: A data acquisition module, used for acquiring sample data of surface drainage facilities, and constructing the sample data of surface drainage facilities into an initial sample data set; A data acquisition module, used for acquiring sample data of surface drainage facilities, and constructing the sample data of surface drainage facilities into an initial sample data set; A defect classification module is used to obtain a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix to obtain defect analysis data; obtain optimized defect classification based on the defect analysis data, and obtain ground drainage facility data after defect classification according to the optimized defect classification; a data labeling module, used for labeling the ground drainage facility data after defect classification to obtain the labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The database building module is used to integrate the corrected ground drainage facility data with marked boxes to form a municipal drainage facility database.

[0021] By adopting the above technical solution, in the process of building a municipal drainage facility database, a large amount of surface drainage facility sample data can be obtained through on-site actual working condition images or on the Internet. The surface drainage facility data refers to the surface drainage facility image. The collected surface drainage facility data are integrated to form an initial sample data set. The initial sample data set is defect-classified to obtain a real defect type set and a predicted defect type set. The real defect type set refers to a set of facility defect types obtained by defect-classifying surface drainage facility sample data according to traditional industry standards. The predicted defect type set refers to a set of defect type prediction analysis of surface drainage facility sample data. The real defect type set and the predicted defect type set are input into a preset multi-label classification confusion matrix to obtain defect analysis data. The defect analysis data refers to the multi-label classification confusion matrix, which determines the surface drainage facility samples according to the proportion of the number of correctly predicted samples to the total number of samples. The accuracy of defect classification of data is improved, and the defect classification is optimized using the obtained defect analysis data. According to the optimized defect classification, the sample data of ground drainage facilities are defect-classified to obtain the ground drainage facilities data after defect classification, and then the ground drainage facilities data after defect classification are labeled, and the defect analysis model is used to repeatedly perform labeling detection on the labeled ground drainage facilities data to obtain the labeling detection results. The labeling of the ground drainage facilities data is corrected according to the labeling detection results, and the corrected ground drainage facilities data with labeling boxes are integrated to form a municipal drainage facilities database. During the entire database construction process, the defect classification of the ground drainage facilities data is optimized using a multi-label classification confusion matrix, and the defect labeling of the ground drainage facilities data is detected and corrected using a defect analysis model, which improves the municipal drainage facilities database's accuracy in distinguishing facility defect types and improves the confusion of database defect types.

[0022] The third objective of the present application is achieved through the following technical solutions: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method for constructing the municipal drainage facility shortage database are implemented.

[0023] The fourth objective of the present application is achieved through the following technical solutions: A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing the municipal drainage facility shortage database.

[0024] In summary, the present application includes at least one of the following beneficial technical effects: 1. In the process of constructing a municipal drainage facility database, a large amount of surface drainage facility sample data can be obtained through on-site actual working condition images or on the Internet. Surface drainage facility data refers to surface drainage facility images. The collected surface drainage facility data are integrated to form an initial sample data set. The initial sample data set is defect-classified to obtain a real defect type set and a predicted defect type set. The real defect type set refers to a set of facility defect types obtained by defect-classifying surface drainage facility sample data according to traditional industry standards. The predicted defect type set refers to a set of defect type prediction analysis of surface drainage facility sample data. The real defect type set and the predicted defect type set are input into a preset multi-label classification confusion matrix to obtain defect analysis data. The defect analysis data refers to the multi-label classification confusion matrix, which determines the accuracy of defect classification of surface drainage facility sample data according to the proportion of the number of correctly predicted samples to the total number of samples. The defect analysis data obtained is used to optimize the defect classification. According to the optimized defect classification, the surface drainage facility sample data is defect-classified to obtain the surface drainage facility data after defect classification. The surface drainage facility data after defect classification is then labeled. The defect analysis model is used to repeatedly perform labeling detection on the labeled ground drainage facility data to obtain the labeling detection results. The labeling of the ground drainage facility data is corrected according to the labeling detection results. The corrected ground drainage facility data with the labeling box is integrated to form a municipal drainage facility database. During the entire database construction process, the defect classification of the ground drainage facility data is optimized using the multi-label classification confusion matrix, and the defect analysis model is used to detect and correct the defect labeling of the ground drainage facility data, thereby improving the accuracy of the municipal drainage facility database in distinguishing the types of facility defects and improving the confusion of database defect types; 2. In the process of collecting ground drainage facility data, the initial ground drainage facility images can be obtained through crawling through search engines on the Internet through the formulated retrieval rules, or the actual working condition images can be obtained by the staff taking photos on site. The collected initial ground drainage facility images and the actual working condition images on site will contain other debris that affect the classification of facility defects. The preset image screening model is used to filter and denoise the actual working condition images on site and the initial ground drainage facility images to obtain a high-quality initial sample data set to improve the data quality in the database; 3. In the process of optimizing the classification of defects in ground drainage facilities, the detection classification is determined by using the facility defect classification rules of traditional industry standards. The sample data in the initial sample data set is first labeled according to the detection classification to obtain a set of real defect types. The sample data in the initial sample data set is then input into the defect analysis model. The predicted defect type set is obtained through the defect analysis model. The real defect type set is used as a horizontal column and the predicted defect type set is used as a vertical column to enter the multi-label classification confusion matrix. The sample data in the initial sample data set is then input into the multi-label classification confusion matrix to obtain the value of each element in the matrix. Each element value is used as defect analysis data, so as to optimize the defect classification of ground drainage facilities according to the specific situation of the defect analysis data, improve the accuracy of distinguishing facility defect types in the municipal drainage facility database, and improve the confusion of database defect types. 4. After inputting the sample data in the initial sample data set into the multi-label classification confusion matrix, extract the values ​​of each element in the multi-label classification confusion matrix to obtain defect analysis data. The defect types that are easily confused in the defect classification are determined through the defect analysis data. By analyzing the easily confused defect types and considering the actual application scenarios of operation and maintenance operations, the defect types in the horizontal and vertical columns of the multi-label classification confusion matrix are optimized according to the easily confused defect types to obtain the optimized multi-label classification confusion matrix. The sample data is then input into the optimized multi-label classification confusion matrix to obtain the ground drainage facility data with good defect classification, which can effectively reduce the intra-class differences and inter-class similarity problems of the data, and provide important support for building a high-quality ground drainage facility database; 5. Perform labeling detection on the labeled ground drainage facility data through the defect analysis model to obtain the labeling detection result. When the labeling detection result is normal labeling, it means that the defect labeling in the ground drainage facility data after defect classification is accurate, and the labeled ground drainage facility data is retained; when the labeling detection result is abnormal labeling, analyze the labeling abnormality to obtain labeling abnormality information. The labeling abnormality information refers to the specific labeling problems that appear in the ground drainage facility data, such as missing labeling, incorrect type labeling, inaccurate labeling range box, over-labeling, difficulty in distinguishing the defect range, etc. According to the labeling abnormality information, correct the labeling box, re-label the ground drainage facility data after defect classification, and repeat the inspection until the labeling result is normal labeling, improve the process of labeling the ground drainage facility data after defect classification, reduce the probability of labeling abnormalities, and improve the quality of the database and the detection accuracy of data labeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 2It is a flowchart for implementing step S10 in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 3 It is a flowchart for implementing step S20 in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 4 It is a flowchart for implementing step S30 in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 5 It is a flowchart for implementing step S40 in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 6 This is another implementation flow chart of step S40 in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 7 It is another implementation flow chart of a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Figure 8 It is a principle block diagram of a system for constructing a municipal drainage facility shortage database in one embodiment of the present application; Fig. 9 is a schematic diagram of a computer device in an embodiment of the present application; Fig.10 It is a multi-label classification confusion matrix diagram in a method for constructing a municipal drainage facility shortage database in one embodiment of the present application; Fig.11 It is a multi-label classification confusion matrix diagram after defect classification optimization in a method for constructing a municipal drainage facility defect database in one embodiment of the present application. DETAILED DESCRIPTION

[0026] The present application is further described in detail below in conjunction with the accompanying drawings.

[0027] In one embodiment, if Figure 1 As shown, the present application discloses a method for constructing a municipal drainage facility shortage database, which specifically includes the following steps: S10: Acquire surface drainage facility sample data, and construct the surface drainage facility sample data into an initial sample data set.

[0028] In this embodiment, the ground drainage facility data refers to the ground drainage facility image.

[0029] Specifically, in the process of constructing a municipal drainage facility database, a large amount of surface drainage facility sample data can be obtained through on-site actual working condition images or on the Internet, and the collected surface drainage facility data can be integrated to form an initial sample data set.

[0030] S20: Obtain a real defect type set and a predicted defect type set according to the initial sample data set, input the real defect type set and the predicted defect type set into a preset multi-label classification confusion matrix to obtain defect analysis data.

[0031] In this embodiment, the real defect type set refers to the set of facility defect types obtained by classifying the sample data of ground drainage facilities according to traditional industry standards, the predicted defect type set refers to the set of defect type prediction analysis of the sample data of ground drainage facilities, and the defect analysis data refers to the proportion of the number of correctly predicted samples to the total number of samples in the multi-label classification confusion matrix.

[0032] Specifically, defect classification is performed on the initial sample data set to obtain a set of facility defect types obtained by classifying the surface drainage facility sample data according to traditional industry standards and a set of defect type prediction analysis of the surface drainage facility sample data. The set of real defect types and the set of predicted defect types are input into a preset multi-label classification confusion matrix to obtain a multi-label classification confusion matrix. The accuracy of defect classification of the surface drainage facility sample data is determined based on the ratio of the predicted defect type to the real defect type.

[0033] S30: Obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification.

[0034] Specifically, the defect analysis data obtained is used to optimize the defect classification, and the defect classification is performed on the surface drainage facility sample data according to the optimized defect classification to obtain the surface drainage facility data after defect classification.

[0035] S40: Label the defect-classified ground drainage facility data to obtain labeled ground drainage facility data, input the labeled ground drainage facility data into a preset defect analysis model, obtain defect labeling detection results, and correct the labeled ground drainage facility data according to the labeling detection results.

[0036] Specifically, the ground drainage facility data after defect classification is labeled, and the labeled ground drainage facility data is repeatedly labeled and detected using the defect analysis model to obtain the labeling detection results, and the labeling of the ground drainage facility data is corrected according to the labeling detection results.

[0037] S50: Integrate the corrected ground drainage facility data with the marked boxes to form a municipal drainage facility database.

[0038] Specifically, the corrected ground drainage facility data with marked boxes are integrated to form a municipal drainage facility database.

[0039] In this embodiment, in the process of constructing a municipal drainage facility database, a large amount of surface drainage facility sample data can be obtained through on-site actual working condition images or on the Internet. The surface drainage facility data refers to the surface drainage facility images. The collected surface drainage facility data are integrated to form an initial sample data set. The initial sample data set is defect-classified to obtain a real defect type set and a predicted defect type set. The real defect type set refers to a set of facility defect types obtained by defect-classifying the surface drainage facility sample data according to traditional industry standards. The predicted defect type set refers to a set of defect type prediction analysis of the surface drainage facility sample data. The real defect type set and the predicted defect type set are input into a preset multi-label classification confusion matrix to obtain defect analysis data. The defect analysis data refers to the multi-label classification confusion matrix, which determines the surface drainage facility sample data according to the proportion of the number of correctly predicted samples to the total number of samples. The accuracy of defect classification is improved, and the defect analysis data is used to optimize the defect classification. According to the optimized defect classification, the sample data of ground drainage facilities are defect-classified to obtain the ground drainage facilities data after defect classification. The ground drainage facilities data after defect classification are labeled, and the defect analysis model is used to repeatedly perform labeling detection on the labeled ground drainage facilities data to obtain the labeling detection results. The labeling of the ground drainage facilities data is corrected according to the labeling detection results, and the corrected ground drainage facilities data with labeling boxes are integrated to form a municipal drainage facilities database. During the entire database construction process, the multi-label classification confusion matrix is ​​used to optimize the defect classification of the ground drainage facilities data, and the defect analysis model is used to detect and correct the defect labeling of the ground drainage facilities data, which improves the municipal drainage facilities database The accuracy of distinguishing facility defect types and improves the confusion of database defect types.

[0040] In one embodiment, if Figure 2 As shown, in step S10, the surface drainage facility sample data is obtained, and the surface drainage facility sample data is constructed into an initial sample data set, which specifically includes: S11: Based on the preset retrieval rules, crawl the initial ground drainage facility images on the Internet.

[0041] S12: Acquire an image of actual working conditions on site, and input the image of actual working conditions on site and an initial surface drainage facility image into a preset image screening model to obtain initial sample data.

[0042] Specifically, in the process of collecting ground drainage facility data, the initial ground drainage facility images can be obtained through search engine crawling on the Internet through the established retrieval rules, or the actual working condition images can be obtained by the staff taking pictures on site. The collected initial ground drainage facility images and the actual working condition images on site will contain other debris that affect the classification of facility defects. The preset image screening model is used to filter and denoise the actual working condition images on site and the initial ground drainage facility images to obtain a high-quality initial sample data set to improve the data quality in the database.

[0043] In one embodiment, if Figure 3 As shown, in step S20, a real defect type set and a predicted defect type set are obtained according to the initial sample data set, and the real defect type set and the predicted defect type set are input into a preset multi-label classification confusion matrix to obtain defect analysis data, which specifically includes: S21: Based on the preset facility defect classification rules, the initial sample data set is defect-classified to obtain a real defect type set.

[0044] S22: Inputting the initial sample data set into a preset defect analysis model to obtain a predicted defect type set.

[0045] S23: Inputting the real defect type set and the predicted defect type set into the multi-label classification confusion matrix, with the real defect type set as the horizontal column and the predicted type set as the vertical column.

[0046] S24: Input the initial sample data into a multi-label classification confusion matrix to obtain defect analysis data.

[0047] Specifically, in the process of optimizing the classification of defects in ground drainage facilities, the detection classification is determined by using the facility defect classification rules of traditional industry standards, and the sample data in the initial sample data set is first labeled according to the detection classification to obtain a set of true defect types, and then the sample data in the initial sample data set is input into the defect analysis model, and the predicted defect type set is obtained through the defect analysis model, and the true defect type set is used as a horizontal column and the predicted defect type set is used as a vertical column to input into a multi-label classification confusion matrix, and then the sample data in the initial sample data set is input into the multi-label classification confusion matrix to obtain the value of each element in the matrix, and each element value is used as defect analysis data, such as in the matrix is the number of samples that are actually classified as class j and predicted as class i. From the perspective of classification quality, the accuracy of sample classification and the classification of samples that are easily confused can be determined by the diagonal and off-diagonal element values ​​in the confusion matrix. For example: m iiThe larger the value is, the closer the predicted classification is to the true classification label, and the higher the classification accuracy is. ik The larger the value, the easier it is to misidentify the i-classification as the k-classification, and the i-classification and the k-classification are easy to be confused, so as to optimize the defect classification of the ground drainage facilities according to the specific situation of the defect analysis data. In one embodiment, if Figure 4 As shown, in step S30, obtaining optimized defect classification based on the defect analysis data, and obtaining defect classified ground drainage facility data according to the optimized defect classification specifically includes: S31: Obtaining easily confused defect classes according to the defect analysis data, and optimizing the multi-label classification confusion matrix based on the easily confused defect classes to obtain an optimized multi-label classification confusion matrix.

[0048] S32: Obtaining defect-classified ground drainage facility data based on the optimized multi-label classification confusion matrix.

[0049] Specifically, after the sample data in the initial sample data set is input into the multi-label classification confusion matrix, the value of each element in the multi-label classification confusion matrix is ​​extracted to obtain defect analysis data. The defect types that are easily confused in the defect classification are determined through the defect analysis data. By analyzing the easily confused defect types and considering the actual application scenarios of operation and maintenance operations, the defect types in the horizontal and vertical columns of the multi-label classification confusion matrix are optimized according to the easily confused defect classes to obtain the optimized multi-label classification confusion matrix. The sample data is then input into the optimized multi-label classification confusion matrix to obtain ground drainage facility data with good defect classification, which can effectively reduce the intra-class differences and inter-class similarity problems of the data, and provide important support for building a high-quality ground drainage facility database. Fig.10 As shown in the figure, there is a high type confusion ratio for the defects of hole blockage (Comb_KYDS) and rain grate blockage (Comb_YBNDS), with 0.12% of the hole blockage being mistakenly identified as rain grate blockage, and 0.15% of the rain grate blockage being mistakenly identified as hole blockage; in order to reduce the intra-class differences and inter-class similarities of defects, and considering the actual application scenarios of drainage facility operation and maintenance operations, these two types of defects have similar effects on the operation of drainage facilities and are handled by manual or mechanical dredging to dredge the rainwater grates. The hole blockage (Comb_KYDS) and rain grate blockage (Comb_YBNDS) are merged into hole blockage (Comb_KYDS), and the leaf morphological characteristics are considered to be obviously different from other types of blockages in the rain grate blockage defect. Based on the existence of leaf occlusion, the presence of leaf occlusion is separated as a new defect label, defined as leaf occlusion (Comb_SYZD). The optimized multi-label classification confusion matrix is ​​as follows Fig.11As shown in the figure, after the classification was reclassified, the classification confusion test was conducted again, and the confusion of defect types was improved, reducing the high confusion between different defect types.

[0050] In one embodiment, if Figure 5 As shown, in step S40, the ground drainage facility data after defect classification is labeled to obtain the labeled ground drainage facility data, which specifically includes: S41: Obtain defect type information based on the ground drainage facility data after defect classification, and obtain defect morphological characteristics and defect range according to the defect type information.

[0051] S42: Generate a labeling frame based on the defect morphological characteristics and defect range, label the ground drainage facility data after defect classification, and obtain labeled ground drainage facility data.

[0052] Specifically, after completing the defect classification of the ground drainage facility data, the corresponding defect type information is obtained according to the classified ground drainage facility data, the defect morphological characteristics and defect range of the ground drainage facility are obtained in the defect type information, and a labeling box is generated according to the defect morphological characteristics and defect range to label the defects of the ground drainage facility data, and the labeled ground drainage facility data is obtained to highlight the defect situation of the ground drainage facility.

[0053] In one embodiment, if Figure 6 As shown, in step S40, the annotated ground drainage facility data is input into a preset defect analysis model to obtain defect marking detection results, and the annotated ground drainage facility data is corrected according to the marking detection results, specifically including: S43: Input the annotated ground drainage facility data into a preset defect analysis model to obtain defect annotation detection results, wherein the annotation detection results include abnormal annotations and normal annotations.

[0054] S44: When the marking detection result is a normal marking, retaining the marked ground drainage facility data.

[0055] S45: When the labeling detection result is an abnormal labeling, obtain labeling abnormality information based on the labeled ground drainage facility data, correct the labeling box according to the labeling abnormality information, re-label the ground drainage facility data after the defect classification, and obtain the labeled ground drainage facility data.

[0056] Specifically, the labeled ground drainage facility data is labeled and detected through the defect analysis model to obtain the labeling detection result. When the labeling detection result is normal labeling, it means that the defect labeling in the ground drainage facility data after defect classification is accurate, and the labeled ground drainage facility data is retained; when the labeling detection result is abnormal labeling, the labeling abnormality is analyzed to obtain labeling abnormality information. The labeling abnormality information refers to specific labeling problems that occur in the ground drainage facility data, such as missing labeling, type labeling error, inaccurate labeling range box, over-labeling, difficulty in distinguishing the defect range, etc. The labeling box is corrected according to the labeling abnormality information, the ground drainage facility data after defect classification is re-labeled, and the inspection is repeated until the labeling result is normal labeling, so as to improve the process of labeling the ground drainage facility data after defect classification, reduce the probability of labeling abnormalities, and improve the quality of the database and the detection accuracy of data labeling. In one embodiment, if Figure 7 As shown, the method for constructing a municipal drainage facility shortage database also includes the following steps: S501: Performing data enhancement processing on the corrected ground drainage facility data with the marked frame to obtain enhanced ground drainage facility data.

[0057] S502: Integrate the enhanced ground drainage facility data to form a municipal drainage facility database.

[0058] Specifically, by performing data enhancement on the corrected ground drainage facility data with labeled boxes, such as geometric transformation operations such as image rotation, scaling, cropping, flipping, random erasing, and optical transformation operations such as brightness, contrast, and saturation, the data diversity can be increased and the quality of the municipal drainage facility database can be improved without changing the original data.

[0059] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0060] In one embodiment, a system for constructing a municipal drainage facility shortage database is provided, and the system for constructing a municipal drainage facility shortage database corresponds one-to-one to the method for constructing a municipal drainage facility shortage database in the above embodiment. Figure 8 As shown in the figure, the construction system of the municipal drainage facility defect database includes data acquisition module, defect classification module, data annotation module and database construction module. The detailed description of each functional module is as follows: A data acquisition module, used for acquiring sample data of surface drainage facilities, and constructing the sample data of surface drainage facilities into an initial sample data set; A defect classification module is used to obtain a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix to obtain defect analysis data; obtain optimized defect classification based on the defect analysis data, and obtain ground drainage facility data after defect classification according to the optimized defect classification; a data labeling module, used for labeling the ground drainage facility data after defect classification to obtain the labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The database building module is used to integrate the corrected ground drainage facility data with marked boxes to form a municipal drainage facility database.

[0061] The specific definition of the system for constructing the municipal drainage facility lack database can be found in the definition of the method for constructing the municipal drainage facility lack database mentioned above, which will not be repeated here. Each module in the above-mentioned system for constructing the municipal drainage facility lack database can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be 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.

[0062] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface and a database 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 multiple data source tables, a joint data table, a first data feature information and a second data feature information. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for constructing a municipal drainage facility shortage database is implemented.

[0063] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are implemented: A method for constructing a municipal drainage facility deficiency database, the method comprising the steps of: acquiring surface drainage facility sample data, and constructing the surface drainage facility sample data into an initial sample data set; Acquire a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix, and obtain defect analysis data; Obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification; Labeling the defect-classified ground drainage facility data to obtain labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The corrected surface drainage facility data with marked boxes are integrated to form a municipal drainage facility database.

[0064] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: A method for constructing a municipal drainage facility deficiency database, the method comprising the steps of: acquiring surface drainage facility sample data, and constructing the surface drainage facility sample data into an initial sample data set; Acquire a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix, and obtain defect analysis data; Obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification; Labeling the defect-classified ground drainage facility data to obtain labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The corrected surface drainage facility data with marked boxes are integrated to form a municipal drainage facility database.

[0065] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0066] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0067] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for constructing a municipal drainage facility shortage database, characterized in that: The method for constructing the municipal drainage facility shortage database comprises the following steps: Acquire surface drainage facility sample data, and construct the surface drainage facility sample data into an initial sample data set; Acquire a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix, and obtain defect analysis data; Obtaining optimized defect classification based on the defect analysis data, and obtaining ground drainage facility data after defect classification according to the optimized defect classification; Labeling the defect-classified ground drainage facility data to obtain labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The corrected surface drainage facility data with marked boxes are integrated to form a municipal drainage facility database.

2. The method for constructing a municipal drainage facility deficiency database according to claim 1, characterized in that: The acquiring of the surface drainage facility sample data and constructing the surface drainage facility sample data into an initial sample data set specifically includes: Acquire an image of the actual working condition on site, and crawl an initial surface drainage facility image on the Internet based on the image of the actual working condition on site; The actual on-site working condition image and the initial surface drainage facility image are input into a preset image screening model to obtain initial sample data.

3. The method for constructing a municipal drainage facility deficiency database according to claim 1, characterized in that: The step of obtaining a set of real defect types and a set of predicted defect types according to the initial sample data set, and inputting the set of real defect types and the set of predicted defect types into a preset multi-label classification confusion matrix to obtain defect analysis data specifically includes: Based on a preset facility defect classification rule, the initial sample data set is defect-classified to obtain a real defect type set; Inputting the initial sample data set into a preset defect analysis model to obtain a predicted defect type set; Inputting the true defect type set and the predicted defect type set into the multi-label classification confusion matrix, with the true defect type set as the horizontal column and the predicted type set as the vertical column; The initial sample data is input into a multi-label classification confusion matrix to obtain defect analysis data.

4. The method for constructing a municipal drainage facility deficiency database according to claim 1, characterized in that: The step of obtaining optimized defect classification based on the defect analysis data and obtaining defect classified ground drainage facility data according to the optimized defect classification specifically includes: Acquire easily confused defect classes according to the defect analysis data, and optimize the multi-label classification confusion matrix based on the easily confused defect classes to obtain an optimized multi-label classification confusion matrix; The defect-classified ground drainage facility data is obtained based on the optimized multi-label classification confusion matrix.

5. The method for constructing a municipal drainage facility deficiency database according to claim 1, characterized in that: The step of labeling the defect-classified ground drainage facility data to obtain the labeled ground drainage facility data specifically includes: Acquire defect type information based on the defect-classified ground drainage facility data, and acquire defect morphological characteristics and defect range according to the defect type information; A labeling frame is generated based on the defect morphological characteristics and defect range, and the ground drainage facility data after defect classification is labeled to obtain the labeled ground drainage facility data.

6. The method for constructing a municipal drainage facility deficiency database according to claim 5, characterized in that: The step of inputting the annotated ground drainage facility data into a preset defect analysis model, obtaining defect annotation detection results, and correcting the annotated ground drainage facility data according to the annotation detection results specifically includes: Input the annotated surface drainage facility data into a preset defect analysis model to obtain defect annotation detection results, wherein the annotation detection results include abnormal annotations and normal annotations; When the marking detection result is a normal marking, retaining the marked ground drainage facility data; When the labeling detection result is an abnormal labeling, labeling abnormality information is obtained based on the labeled ground drainage facility data, the labeling box is corrected according to the labeling abnormality information, and the ground drainage facility data after the defect classification is re-labeled to obtain the labeled ground drainage facility data.

7. The method for constructing a municipal drainage facility deficiency database according to claim 1, characterized in that: Before the modified ground drainage facility data with the marked frame is integrated to form a municipal drainage facility database, the method for constructing a municipal drainage facility database further includes the steps of: Performing data enhancement processing on the corrected ground drainage facility data with the marked frame to obtain enhanced ground drainage facility data; The enhanced surface drainage facility data are integrated to form a municipal drainage facility database.

8. A system for constructing a municipal drainage facility shortage database, characterized in that: The system for constructing the municipal drainage facility shortage database includes: A data acquisition module, used for acquiring sample data of surface drainage facilities, and constructing the sample data of surface drainage facilities into an initial sample data set; A defect classification module is used to obtain a real defect type set and a predicted defect type set according to the initial sample data set, construct a multi-label classification confusion matrix based on the real defect type set and the predicted defect type set, input the initial sample data set into the multi-label classification confusion matrix to obtain defect analysis data; obtain optimized defect classification based on the defect analysis data, and obtain ground drainage facility data after defect classification according to the optimized defect classification; a data labeling module, used for labeling the ground drainage facility data after defect classification to obtain the labeled ground drainage facility data, inputting the labeled ground drainage facility data into a preset defect analysis model, repeatedly detecting the labeled ground drainage facility data to obtain a defect labeling detection result, and correcting the labeled ground drainage facility data according to the labeling detection result; The database building module is used to integrate the corrected ground drainage facility data with marked boxes to form a municipal drainage facility database.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the processor implements the steps of a method for constructing a municipal drainage facility shortage database as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a method for constructing a municipal drainage facility shortage database as described in any one of claims 1 to 7 are implemented.

Citation Information

Patent Citations

  • Defect classification model generation method and device, classification method and terminal equipment

    CN115147335A

  • Model-based defect detection method and system

    CN115495608A

  • Domain adaptation metal surface defect detection method oriented to small amount of target data annotation

    CN117611533A

  • Sewer line defect detection method and system based on confusion matrix

    CN117893818A

  • A method and device for classifying defects in drone inspection based on the self-distillation of similarity consistency

    DE102023125558A1