Earthquake damage structure damage segmentation method and device, electronic equipment and storage medium

Through the construction and training of deep learning models, the accuracy and efficiency of post-seismic concrete structure damage segmentation are solved, and the rapid positioning and accurate segmentation of seismic RC structures are achieved, which improves the efficiency of post-seismic evaluation.

CN119048591AActive Publication Date: 2024-11-29XI'AN UNIVERSITY OF ARCHITECTURE AND TECHNOLOGY

Patent Information

Application Number
CN202410992162.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-29
Estimated Expiration
2044-07-23

AI Technical Summary

Technical Problem

The existing technology cannot accurately identify and segment seismic damage of concrete structures, especially in post-seismic evaluation, which cannot meet the quantitative requirements of damage degree. The application of computer vision in the field of civil engineering is mainly limited to crack identification, and there is a lack of multi-category damage segmentation model. .

Method used

A method of damage segmentation of seismic loss structures is proposed. By acquiring and labeling original pictures, data enhancement is carried out, and a deep learning model of the U-shaped jump link network architecture of pyramid multi-scale feature fusion, Haar wavelet downsampling and attention module is used to Multi-category cross-entropy loss function with weights is trained to realize damage segmentation.

Benefits of technology

It realizes rapid positioning and accurate segmentation of visible seismic damage to seismic RC structures, improves the efficiency of post-seismic evaluation work, overcomes the problems of insufficient data sets and insufficient model applicability, and has high practical engineering significance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048591A_ABST
    Figure CN119048591A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of concrete structure earthquake damage identification, in particular to an earthquake damage structure damage segmentation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining an original image of an earthquake damage RC structure, and generating a training data set; labeling the training data set to obtain a labeled data set; performing data enhancement on the labeled data set to obtain an enhanced data set; constructing an earthquake damage structure damage segmentation model, and training the earthquake damage structure damage segmentation model by using the enhanced data set to obtain a trained earthquake damage structure damage segmentation model; and performing damage segmentation on the to-be-segmented image by using the trained seismic damage structure damage segmentation model to obtain a segmentation result. A multi-class cross entropy loss function with a weight is used, a learning rate scheduler adopts a self-adaptive learning rate automatic scheduler, the efficiency of post-earthquake evaluation work is greatly improved, and the established model is simple in architecture, small in parameter quantity, suitable for local deployment and high in practical engineering significance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of earthquake damage identification of concrete structures, and in particular to a method, device, electronic device and storage medium for segmenting earthquake-damaged structure damage. Background Art

[0002] Accurate visual inspection of earthquake-damaged RC structures is a tedious task and places high demands on the professionalism of the assessor. In addition, different assessors may give different assessment results for the same damaged component. These problems limit the efficiency and accuracy of existing post-earthquake assessment work.

[0003] In recent years, the rapid development of computer vision technology has provided many new paradigms for many problems in the engineering field, and has great development potential in the field of post-earthquake assessment. However, the current application of computer vision in the field of civil engineering is still limited to early structural health detection based on crack identification. Unlike health monitoring research, post-earthquake assessment of RC structures requires people to accurately identify more extensive visible visual damage information such as concrete spalling and exposed steel bars. Although there are some damage location models for earthquake-damaged RC structures based on target detection methods, these models can only locate the damage but cannot segment the damage, and cannot meet the needs of quantifying the degree of damage in post-earthquake assessment. However, due to the lack of public data sets, there are currently very few models that can be used for multi-category damage segmentation of earthquake-damaged RC structures. Summary of the invention

[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the first purpose of this application is to propose a method for segmenting damage to earthquake-damaged structures to solve the problems that existing technical means cannot achieve damage segmentation and cannot meet the needs of quantifying the degree of damage in post-earthquake assessment.

[0006] The second object of the present application is to provide a device.

[0007] The third objective of the present application is to provide an electronic device.

[0008] A fourth objective of the present application is to provide a computer-readable storage medium.

[0009] To achieve the above-mentioned purpose, the first embodiment of the present application proposes a method for segmenting damage of a seismic damaged structure, comprising:

[0010] Obtain original images of damaged RC structures and generate training data sets;

[0011] Performing labeling processing on the training data set to obtain a labeled data set;

[0012] Performing data enhancement on the labeled data set to obtain an enhanced data set;

[0013] Constructing a damage segmentation model for a seismic-damaged structure, and using the enhanced data set to train the damage segmentation model for the seismic-damaged structure to obtain a trained damage segmentation model for the seismic-damaged structure;

[0014] The trained earthquake-damaged structure damage segmentation model is used to perform damage segmentation on the image to be segmented to obtain a segmentation result.

[0015] Preferably, the obtaining of original images of the damaged RC structure and generating a training data set comprises:

[0016] The training data set is generated by obtaining test pictures taken during RC beam, column and node tests, illustrations of papers on RC component test research, actual damage photos of RC structures damaged in earthquake-stricken areas, and images of RC components taken in different environments and backgrounds.

[0017] Preferably, the step of labeling the training data set to obtain the labeled data set includes:

[0018] The pixels in the training data set are labeled, including background, concrete spalling, concrete cracking, concrete crushing, exposed longitudinal reinforcement, exposed longitudinal reinforcement, buckling longitudinal reinforcement, exposed stirrups, and concrete fragments. After labeling, each pixel point in the feature map exported has a unique category, generating a labeled data set.

[0019] Preferably, performing data enhancement on the annotated data set to obtain the enhanced data set includes: performing data enhancement processing on the annotated data set by using mirror flipping, color distortion and random Gaussian noise to obtain the enhanced data set.

[0020] Preferably, constructing the damage segmentation model of the earthquake-damaged structure includes:

[0021] A damage segmentation model for earthquake-damaged structures is constructed using a U-shaped network architecture with skip links using pyramid multi-scale feature fusion, Haar wavelet downsampling, and attention modules.

[0022] Preferably, the using the enhanced data set to train the earthquake-damaged structure damage segmentation model to obtain the trained earthquake-damaged structure damage segmentation model comprises:

[0023] An adaptive learning rate automatic scheduler is used to dynamically adjust the learning rate according to the change of loss during training, and weighted multi-category cross entropy is used as the loss function. When the loss change is small, that is, the model tends to be stable, the learning rate is reduced for more precise adjustment; when the loss change is large, the learning rate is increased to accelerate convergence.

[0024] Preferably, the multi-category cross entropy loss function calculation formula is:

[0025]

[0026] Among them, E is the weighted cross entropy loss, α is the weight, k is the total category, α s is the scaled weight, t k is the label, y k is the predicted value.

[0027] To achieve the above-mentioned purpose, the second embodiment of the present application proposes a device for segmenting damage of earthquake-damaged structures, comprising:

[0028] Data acquisition module, which obtains original images of earthquake-damaged RC structures and generates training data sets;

[0029] A data annotation module is used to annotate the training data set to obtain an annotated data set;

[0030] A data enhancement module performs data enhancement on the labeled data set to obtain an enhanced data set;

[0031] A model training module is used to construct a damage segmentation model for earthquake-damaged structures, and to train the damage segmentation model for earthquake-damaged structures using the enhanced data set to obtain a trained damage segmentation model for earthquake-damaged structures;

[0032] The model prediction module uses the trained earthquake-damaged structure damage segmentation model to perform damage segmentation on the image to be segmented to obtain a segmentation result.

[0033] To achieve the above-mentioned purpose, the third aspect of the present application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0034] The memory stores computer-executable instructions;

[0035] The processor executes the computer-executable instructions stored in the memory to implement any of the above methods.

[0036] To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, including computer-executable instructions stored in the computer-readable storage medium, and the computer-executable instructions are used to implement any of the methods described above when executed by a processor.

[0037] A method for segmenting earthquake-damaged structure damage provided in the present application overcomes the defect of insufficient public data sets in the earthquake-damaged RC structure damage segmentation task based on computer vision, and greatly improves the efficiency of post-earthquake assessment work. The deep learning model proposed in the present application has small structural parameters, and the established model architecture is concise, with a small number of parameters, which is suitable for local deployment, and can achieve rapid positioning and accurate segmentation of visible earthquake damage to earthquake-damaged RC structures, thereby improving the efficiency of post-earthquake assessment work and having high practical engineering significance.

[0038] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0040] Figure 1 A flowchart of a first specific embodiment of a method for segmenting damage of a structure damaged by an earthquake provided by the present invention;

[0041] Figure 2 Some pictures in the dataset;

[0042] Figure 3 It is the deep learning network model architecture;

[0043] Figure 4 A structural block diagram of a device for segmenting damage of earthquake-damaged structures provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The core of the present invention is to provide a method, device, electronic device and storage medium for segmenting damage of earthquake-damaged structures. Through a deep learning model, the rapid positioning and accurate segmentation of visible earthquake damage to earthquake-damaged RC structures can be achieved, thereby improving the efficiency of post-earthquake assessment work.

[0045] In order to enable those skilled in the art to better understand the scheme of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0046] Please refer to Figure 1 , Figure 1 This is a flow chart of a first specific embodiment of a method for segmenting damage of earthquake-damaged structures provided by the present invention; the specific operation steps are as follows:

[0047] Step S101: obtaining original images of earthquake-damaged RC structures and generating a training data set;

[0048] The training data set is generated by obtaining test pictures taken during RC beam, column and node tests, illustrations of papers on RC component test research, actual damage photos of RC structures damaged in earthquake-stricken areas, and images of RC components taken in different environments and backgrounds.

[0049] Step S102: labeling the training data set to obtain a labeled data set;

[0050] The pixels in the training data set are labeled, including background, concrete spalling, concrete cracking, concrete crushing, exposed longitudinal reinforcement, exposed longitudinal reinforcement, buckling longitudinal reinforcement, exposed stirrups, and concrete fragments. After labeling, each pixel point in the feature map exported has a unique category, generating a labeled data set.

[0051] Step S103: performing data enhancement on the annotated data set to obtain an enhanced data set;

[0052] The labeled data set is enhanced by using mirror flipping, color distortion and random Gaussian noise to obtain an enhanced data set.

[0053] Step S104: constructing a damage segmentation model for a seismic-damaged structure, and using the enhanced data set to train the damage segmentation model for the seismic-damaged structure to obtain a trained damage segmentation model for the seismic-damaged structure;

[0054] A damage segmentation model for earthquake-damaged structures is constructed using a U-shaped network architecture with skip links using pyramid multi-scale feature fusion, Haar wavelet downsampling, and attention modules.

[0055] An adaptive learning rate automatic scheduler is used to dynamically adjust the learning rate according to the change of loss during training, and weighted multi-category cross entropy is used as the loss function. When the loss change is small, that is, the model tends to be stable, the learning rate is reduced for more precise adjustment; when the loss change is large, the learning rate is increased to accelerate convergence.

[0056] The multi-category cross entropy loss function calculation formula is:

[0057]

[0058] Among them, E is the weighted cross entropy loss, α is the weight, k is the total category, α s is the scaled weight, t k is the label, u k is the predicted value.

[0059] Step S105: using the trained earthquake-damaged structure damage segmentation model to perform damage segmentation on the image to be segmented to obtain a segmentation result.

[0060] A damage segmentation model for earthquake-damaged structures is constructed using a U-shaped network architecture with skip links using pyramid multi-scale feature fusion, Haar wavelet downsampling, and attention modules.

[0061] This embodiment provides a method for segmenting damage of earthquake-damaged structures, which overcomes the defect of insufficient public data sets in the task of segmenting damage of earthquake-damaged RC structures based on computer vision, and greatly improves the efficiency of post-earthquake assessment work. The deep learning model proposed in this application has small structural parameters, and the established model architecture is concise, with a small number of parameters, which is suitable for local deployment, and can achieve rapid positioning and accurate segmentation of visible earthquake damage to earthquake-damaged RC structures, thereby improving the efficiency of post-earthquake assessment work and having high practical engineering significance.

[0062] Based on the above embodiment, this embodiment describes a method for segmenting damage of a seismic damaged structure, which is specifically as follows:

[0063] The purpose of this application is to propose a deep learning model and training data set construction method for damage segmentation of earthquake-damaged RC structures, so as to improve the efficiency of post-earthquake assessment work and improve the number and quality of input samples of earthquake-damaged RC structure damage identification model based on computer vision. The constructed data set contains 1,400 pictures of different structural component types and different degrees of damage. The different structural damage categories in the pictures have been manually annotated at the pixel level. The proposed deep learning model for damage segmentation includes multi-scale feature fusion, Haar wavelet downsampling and attention module to solve the challenges of small sample size, low image resolution and complex background environment in earthquake damage segmentation tasks.

[0064] The original images of earthquake-damaged RC structures that can be used to construct the model training dataset are obtained and screened from existing RC structure test studies and publicly available image datasets; the image sources mainly include the following sources: test images taken in the tests of RC beams, columns, nodes, etc. conducted by the applicant's research team in recent years; illustrations in existing RC component test studies; actual damage images of earthquake-damaged RC structures in actual earthquake-stricken areas shown in existing earthquake damage reports; search engines (e.g., Google Image). The images in the dataset mainly include various types of RC structural components (beams, columns, nodes, walls, etc.). The images in the dataset were taken in many different environments (laboratories, environments with dim light, earthquake-stricken areas, etc.), and contain various types of complex background information and random occlusions (plant occlusions, concrete fragments, etc.).

[0065] Manual pixel-level annotation is performed on different damage categories in the collected images to obtain the damage feature map corresponding to the image; Figure 2 As shown in the figure, there are 8 types of pixels labeled in the picture, namely background, concrete spalling, concrete cracking, concrete crushing, longitudinal reinforcement leakage, longitudinal reinforcement buckling, stirrup leakage, and concrete fragments. The manual labeling process is based on the labelme labeling tool. After labeling, each pixel in the exported feature map has a unique category.

[0066] Perform local data enhancement on the original image and the annotated damage feature map. Three data enhancement methods are applied to each image, namely mirror flipping, color distortion, and random Gaussian noise.

[0067] The data-enhanced image dataset is input into the deep learning network for training, such as Figure 3 As shown, the model network structure used is a U-shaped network architecture with skip links that includes pyramid multi-scale feature fusion, Haar wavelet downsampling, and attention modules;

[0068] The loss function used in training is a weighted multi-category cross entropy loss function, and the learning rate scheduler adopts the ALRS learning rate automatic scheduler; the loss function used in training is a weighted multi-category cross entropy loss function, and its calculation formula is:

[0069]

[0070] Among them, E is the weighted cross entropy loss, α is the weight, k is the total category, α s is the scaled weight, t k is the label, y k is the predicted value.

[0071] A method for segmenting earthquake-damaged structures provided by an embodiment of the present invention overcomes the defect of insufficient public data sets in earthquake-damaged RC structure damage segmentation tasks based on computer vision. In addition, it also overcomes the defect that existing semantic segmentation models are not applicable to earthquake-damaged RC structure damage segmentation tasks. The constructed data set includes various types of RC structural components (beams, columns, nodes, walls, etc.). The pictures in the data set were taken in many different environments (laboratories, environments with dim light, earthquake-stricken areas, etc.), and contain various types of complex background information and random occlusions (plant occlusions, concrete fragments, etc.), which are suitable for various computer vision tasks. The deep learning network structure proposed in the present invention has a small number of parameters, which is conducive to the rapid deployment of the model on local detection equipment, and is convenient for promotion and use in actual engineering. In general, the present invention can achieve rapid positioning and accurate segmentation of visible earthquake damage to earthquake-damaged RC structures, and improve the efficiency of post-earthquake assessment work.

[0072] Please refer to Figure 4 , Figure 4 A structural block diagram of a device for segmenting damage of earthquake-damaged structures provided in an embodiment of the present invention; the specific device may include:

[0073] The data acquisition module 100 acquires the original image of the damaged RC structure and generates a training data set;

[0074] The data annotation module 200 performs annotation processing on the training data set to obtain an annotated data set;

[0075] A data enhancement module 300 performs data enhancement on the labeled data set to obtain an enhanced data set;

[0076] A model training module 400 is used to construct a damage segmentation model for a seismic damaged structure, and to train the damage segmentation model for the seismic damaged structure using the enhanced data set to obtain a trained damage segmentation model for the seismic damaged structure;

[0077] The model prediction module 500 uses the trained earthquake-damaged structure damage segmentation model to perform damage segmentation on the image to be segmented to obtain a segmentation result.

[0078] A device for segmenting damage to a structure damaged by an earthquake in this embodiment is used to implement the aforementioned method for segmenting damage to a structure damaged by an earthquake. Therefore, the specific implementation method of a device for segmenting damage to a structure damaged by an earthquake can be seen in the embodiment part of the method for segmenting damage to a structure damaged by an earthquake in the preceding text. For example, the data acquisition module 100, the data annotation module 200, the data enhancement module 300, the model training module 400, and the model prediction module 500 are respectively used to implement steps S101, S102, S103, S104, and S105 in the aforementioned method for segmenting damage to a structure damaged by an earthquake. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part, which will not be repeated here.

[0079] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0080] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0081] In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0082] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0083] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0084] The present application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, the present disclosure is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0085] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0086] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0087] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0088] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0089] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0090] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0092] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A damage segmentation method for earthquake-damaged structures. It is characterized in that include: Obtain original images of damaged RC structures and generate training data sets; Performing labeling processing on the training data set to obtain a labeled data set; Performing data enhancement on the labeled data set to obtain an enhanced data set; Constructing a damage segmentation model for a seismic-damaged structure, and using the enhanced data set to train the damage segmentation model for the seismic-damaged structure to obtain a trained damage segmentation model for the seismic-damaged structure; The trained earthquake-damaged structure damage segmentation model is used to perform damage segmentation on the image to be segmented to obtain a segmentation result.

2. The damage segmentation method for earthquake-damaged structures according to claim 1, It is characterized in that The obtaining of the original image of the damaged RC structure and generating the training data set comprises: The training data set is generated by obtaining test pictures taken during RC beam, column and node tests, illustrations of papers on RC component test research, actual damage photos of RC structures damaged in earthquake-stricken areas, and images of RC components taken in different environments and backgrounds.

3. The damage segmentation method for earthquake-damaged structures according to claim 1, It is characterized in that The labeling of the training data set to obtain the labeled data set includes: The pixels in the training data set are labeled, including background, concrete spalling, concrete cracking, concrete crushing, exposed longitudinal reinforcement, exposed longitudinal reinforcement, buckling longitudinal reinforcement, exposed stirrups, and concrete fragments. After labeling, each pixel point in the feature map exported has a unique category, generating a labeled data set.

4. The damage segmentation method for earthquake-damaged structures according to claim 1, It is characterized in that The step of performing data enhancement on the annotated data set to obtain an enhanced data set includes: performing data enhancement processing on the annotated data set by using mirror flipping, color distortion and random Gaussian noise to obtain an enhanced data set.

5. The damage segmentation method for earthquake-damaged structures according to claim 1, It is characterized in that The construction of the earthquake-damaged structure damage segmentation model comprises: A damage segmentation model for earthquake-damaged structures is constructed using a U-shaped network architecture with skip links using pyramid multi-scale feature fusion, Haar wavelet downsampling, and attention modules.

6. The damage segmentation method for earthquake-damaged structures according to claim 5, It is characterized in that The method of training the earthquake-damaged structure damage segmentation model by using the enhanced data set to obtain the trained earthquake-damaged structure damage segmentation model includes: An adaptive learning rate automatic scheduler is used to dynamically adjust the learning rate according to the change of loss during training, and weighted multi-category cross entropy is used as the loss function. When the loss change is small, that is, the model tends to be stable, the learning rate is reduced for more precise adjustment; when the loss change is large, the learning rate is increased to accelerate convergence.

7. The method for segmenting earthquake-damaged structures according to claim 6, It is characterized in that The multi-category cross entropy loss function calculation formula is: Among them, E is the weighted cross entropy loss, α is the weight, k is the total category, α s is the scaled weight, t k is the label, u k is the predicted value.

8. A device for segmenting damage to earthquake-damaged structures, It is characterized in that include: Data acquisition module, which obtains original images of earthquake-damaged RC structures and generates training data sets; A data annotation module is used to annotate the training data set to obtain an annotated data set; A data enhancement module performs data enhancement on the labeled data set to obtain an enhanced data set; A model training module is used to construct a damage segmentation model for earthquake-damaged structures, and to train the damage segmentation model for earthquake-damaged structures using the enhanced data set to obtain a trained damage segmentation model for earthquake-damaged structures; The model prediction module uses the trained earthquake-damaged structure damage segmentation model to perform damage segmentation on the image to be segmented to obtain a segmentation result.

9. An electronic device, It is characterized in that include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Citation Information

Patent Citations

  • Building earthquake damage and residual capacity evaluation method based on deep machine learning

    CN113868750A

  • Vehicle image recognition model establishment and recognition method

    CN115375959A

  • Blade damage identification method suitable for small sample condition

    CN115457551A

  • Post-fire concrete structure damage segmentation method based on image semantic segmentation

    CN118172667A

  • Image segmentation method and system for pavement disease based on deep learning

    US20210319561A1

Cited By

  • Earthquake damage RC component damage assessment method based on eye movement attention

    CN121904437A