Method for evaluating building disaster damage of remote sensing image based on visual base model

Abstract

This invention discloses a method for assessing building damage from disasters using remote sensing imagery based on a visual fundamental model. The method includes: acquiring pre-disaster and post-disaster remote sensing images of the building disaster area; inputting the pre-disaster remote sensing images into a trained building location feature extraction network to obtain building location results; wherein the building location feature extraction network includes an encoder module, a decoder, and a mask prediction head, and the processing steps of the building location feature extraction network include: inputting the pre-disaster remote sensing images into the encoder module to obtain building location features, inputting the building location features into the decoder to obtain location decoding features, inputting the location decoding features into the mask prediction head to obtain a building mask; and inputting the post-disaster remote sensing images and building location features into a trained damage perception and discrimination network to obtain building damage classification results. This invention can improve the accuracy and stability of building disaster damage assessment.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image recognition, and in particular to a method, system, storage medium, and electronic device for assessing building disaster damage based on a visual model. Background Technology

[0002] Following natural disasters and emergencies, rapidly and accurately obtaining information on building damage is crucial for emergency response, disaster assessment, and post-disaster reconstruction planning. With the continuous development of remote sensing technology and Earth observation methods, high-resolution remote sensing imagery can cover large areas in a short time after a disaster, providing an important data source for building damage assessment. Therefore, automatic building disaster damage assessment technology based on remote sensing imagery has become an important research and application direction in the field of disaster emergency management.

[0003] Currently, technical solutions for assessing building damage in remote sensing imagery mainly rely on deep learning models to perform change analysis or semantic segmentation on pre- and post-disaster images. By judging changes in pixels or regions within the image, the damage level of the building is directly output. While these methods can achieve a degree of automation under ideal conditions, they still face many challenges in real-world disaster scenarios. Firstly, after a disaster, buildings are often accompanied by complex phenomena such as collapse, obstruction, smoke and dust cover, and structural damage, leading to blurred building boundaries, irregular shapes, and significantly increased difficulty in distinguishing between the background and the target. Secondly, remote sensing images from different disaster types, regions, and imaging conditions vary greatly in terms of illumination, resolution, viewing angle, and background complexity, limiting the generalization ability of traditional models trained on specific datasets or single disaster scenarios.

[0004] In recent years, general-purpose visual foundation models have demonstrated strong feature representation capabilities and cross-scene adaptability in tasks such as image segmentation and target recognition, providing new technical approaches for intelligent interpretation of remote sensing images. However, most existing general-purpose visual foundation models are trained based on natural scene data. When directly applied to remote sensing disaster images, they still suffer from problems such as significant domain differences and insufficient adaptation to disaster features, making it difficult to meet the accuracy and stability requirements for building disaster damage assessment. Summary of the Invention

[0005] This invention provides a method, system, storage medium, and electronic device for assessing building disaster damage based on a visual model, which can improve the accuracy and stability of building disaster damage assessment.

[0006] This invention provides a method for assessing building damage from remote sensing images based on a visual fundamental model, comprising: Acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The pre-disaster remote sensing image is input into a trained building location feature extraction network to obtain building location results. The building location feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building location feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building location features, inputting the building location features into the decoder to obtain location decoding features, and inputting the location decoding features into the mask prediction head to obtain a building mask. The post-disaster remote sensing images and the building location features are input into a trained damage perception and discrimination network to obtain building damage classification results.

[0007] Furthermore, in the aforementioned method for assessing building disaster damage based on a visual fundamental model using remote sensing imagery, the encoder module includes a ViT image encoder and a CNN image encoder. The pre-disaster remote sensing image is input into the encoder module to obtain building location features, including: The pre-disaster remote sensing images are input into the ViT image encoder to obtain global features; The pre-disaster remote sensing images are input into the CNN image encoder to obtain local texture features; Based on the gating mechanism, the global features and the local texture features are fused to obtain building location features.

[0008] Furthermore, the aforementioned method for assessing building disaster damage based on a visual fundamental model, after obtaining the building mask, further includes: The building location result is obtained based on the spatial position corresponding to the building mask.

[0009] Furthermore, in the aforementioned method for assessing building disaster damage based on a visual fundamental model using remote sensing imagery, the pre-disaster remote sensing imagery is input into the ViT image encoder to obtain global features, which are expressed by the following formula:

[0010] in, These are pre-disaster remote sensing images. For ViT image encoder, For global features; The pre-disaster remote sensing images are input into the CNN image encoder to obtain local texture features, which are expressed by the following formula:

[0011] in, For CNN image encoders, This refers to local texture features.

[0012] Furthermore, in the aforementioned method for assessing building damage based on a visual fundamental model using remote sensing imagery, the damage perception and discrimination network includes a ViT image encoder, a CNN image encoder, a decoder, and a classification head; the processing steps of the damage perception and discrimination network include: The post-disaster remote sensing images are input into the ViT image encoder to obtain texture change features, and the pre-disaster remote sensing images are input into the CNN image encoder to obtain context features. Based on the gating mechanism, the texture change features and the context features are fused to obtain the disaster change features; The damage characteristics are obtained by fusing the disaster change characteristics and the building location characteristics based on the channel attention mechanism and the spatial attention mechanism; The damage features are input into the decoder to obtain the instance query vector, and the instance query vector is input into the classification head to obtain the building damage classification result.

[0013] Furthermore, the aforementioned method for assessing building damage based on a visual fundamental model using remote sensing imagery includes inputting the damage features into a decoder to obtain an instance query vector, and inputting the instance query vector into a classification head to obtain a building damage classification result, including: A set of learnable query vectors is predefined, with each vector representing a potential building instance; The learnable query vector and the damaged features are cross-attention calculated to obtain the instance query vector; The instance query vector is input into the classification header to obtain the building damage classification result.

[0014] Furthermore, the aforementioned method for assessing building disaster damage based on visual fundamental models using remote sensing imagery further includes: The building location feature extraction network is trained using a first loss function, which is:

[0015] in, These are segmentation loss, masking loss, and DICE loss. These are the weights corresponding to each loss; The damage perception and discrimination network is trained using a second loss function, which is:

[0016] in, It is the actual label distribution. It is the probability distribution predicted for each category. These are the loss weights for each category. It is a true category index.

[0017] This invention also provides a remote sensing image-based building disaster damage assessment system based on a visual fundamental model, comprising: The acquisition module is used to acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The positioning module is used to input the pre-disaster remote sensing image into a trained building positioning feature extraction network to obtain building positioning results. The building positioning feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building positioning feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building positioning features; inputting the building positioning features into the decoder to obtain positioning decoding features; and inputting the positioning decoding features into the mask prediction head to obtain a building mask. The classification module is used to input the post-disaster remote sensing images and the building location features into the trained damage perception and discrimination network to obtain the building damage classification results.

[0018] The present invention also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute any of the above-described methods for assessing building disaster damage based on a visual fundamental model using remote sensing imagery.

[0019] The present invention also provides an electronic device, including a processor and a memory, wherein the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used in the steps of the remote sensing image building disaster damage assessment method based on visual basic model described above.

[0020] This invention provides a method, system, storage medium, and electronic device for assessing building disaster damage based on a visual fundamental model using remote sensing images. The invention utilizes pre-disaster and post-disaster dual-temporal remote sensing images. The post-disaster remote sensing image is input into a building location feature extraction network to obtain building location results, thus achieving accurate pre-disaster building target location. Furthermore, the pre-disaster building prior features (i.e., building location features) and post-disaster remote sensing image features are jointly input into a damage perception and discrimination network to obtain building damage classification results, achieving refined discrimination of different damage levels. Attached Figure Description

[0021] The technical solution and other beneficial effects of the present invention will become apparent from the following detailed description of specific embodiments of the invention, in conjunction with the accompanying drawings.

[0022] Figure 1A flowchart of a method for assessing building damage based on a visual model in remote sensing images, provided in an embodiment of the present invention.

[0023] Figure 2 The images show building masks obtained by the method provided in the embodiments of the present invention, building masks obtained by other methods, and comparison images of pre-disaster and post-disaster images.

[0024] Figure 3 This is a schematic diagram of the structure of a remote sensing image-based building disaster damage assessment system provided in an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] To address the aforementioned problems, embodiments of the present invention provide a method, system, storage medium, and electronic device for assessing building disaster damage based on a visual model. The visual model-based remote sensing image building disaster damage assessment system provided by this invention can be integrated into an electronic device, which may be a terminal, server, or other such device. The terminal may include a tablet computer, laptop computer, personal computer (PC), miniature processing box, or other devices.

[0028] Please see Figure 1 , Figure 1 The flowchart illustrates a method for assessing building damage based on a visual model in remote sensing imagery, which is applied in electronic devices. This method includes the following steps: S1 acquires pre-disaster and post-disaster remote sensing images of the building disaster area.

[0029] Furthermore, preprocessing is performed on pre-disaster and post-disaster remote sensing images, including cropping and data augmentation. Data augmentation includes random cropping, scaling, brightness perturbation, and rotation operations.

[0030] S2, input the pre-disaster remote sensing image into the trained building location feature extraction network to obtain the building location result; wherein, the building location feature extraction network includes an encoder module, a decoder and a mask prediction head, and the processing of the building location feature extraction network includes: inputting the pre-disaster remote sensing image into the encoder module to obtain building location features, inputting the building location features into the decoder to obtain location decoding features, and inputting the location decoding features into the mask prediction head to obtain the building mask.

[0031] In one embodiment, a pre-trained visual base model is selected as the feature extraction backbone network. The visual base model is pre-trained based on large-scale general image data, and the model structure adopts a ViT-like Transformer architecture, which can simultaneously model long-distance global dependencies.

[0032] Specifically, the building location feature extraction network encoder module includes a ViT image encoder and a CNN image encoder. Pre-disaster remote sensing images are input into the encoder module to obtain building location features, including the following steps: S21, input the pre-disaster remote sensing image into the ViT image encoder to obtain global features.

[0033] This can be expressed by the following formula:

[0034] in, These are pre-disaster remote sensing images. For ViT image encoder, This is a global feature.

[0035] The ViT image encoder consists of multiple ViT coding blocks. The pre-disaster remote sensing images are input into multiple ViT coding blocks to obtain multiple intermediate features. Then, the intermediate features are fused through linear mapping and scale alignment operations to obtain global features.

[0036] S22, input the pre-disaster remote sensing images into the CNN image encoder to obtain local texture features.

[0037] This can be expressed by the following formula:

[0038] in, For CNN image encoders, This refers to local texture features.

[0039] S23, based on a gating mechanism, fuses global features and local texture features to obtain building location features.

[0040] This can be expressed by the following formula:

[0041] in, It is a dynamic gating fusion weight.

[0042] Figure 2 The comparison images of building masks obtained by the method provided in this embodiment of the invention, building masks obtained by other methods, and pre-disaster and post-disaster images show that the building mask obtained by this method is clearer.

[0043] Furthermore, after obtaining the building mask, the process also includes: obtaining the building location result based on the spatial location corresponding to the building mask.

[0044] S3 inputs post-disaster remote sensing images and building location features into a trained damage perception and discrimination network to obtain building damage classification results.

[0045] In one embodiment, the damage-aware discrimination network includes a ViT image encoder, a CNN image encoder, a decoder, and a classification head; the processing procedure of the damage-aware discrimination network includes: S31. Input the post-disaster remote sensing images into the ViT image encoder to obtain texture change features, and input the pre-disaster remote sensing images into the CNN image encoder to obtain contextual features.

[0046] Specifically, it is expressed by the following formula:

[0047]

[0048] in, These are remote sensing images from after the disaster. For texture variation features, For contextual features.

[0049] S32, based on a gating mechanism, fuses texture change features and contextual features to obtain disaster change features.

[0050] Specifically, it is expressed by the following formula:

[0051] in, It is a dynamic gating fusion weight.

[0052] S33, based on the channel attention mechanism and spatial attention mechanism, fuses disaster change characteristics and building location characteristics to obtain damage characteristics.

[0053] Specifically, the building location features and disaster change features are fused through channel attention and spatial attention to achieve the modeling of changes between the two temporal features and obtain the damage features.

[0054] S34. Input the damage features into the decoder to obtain the instance query vector, and input the instance query vector into the classification head to obtain the building damage classification result.

[0055] Specifically, step S27 includes the following steps: S341, a set of learnable query vectors are preset, each vector representing a potential building instance; S342, perform cross-attention calculation on the learnable query vector and the damaged features to obtain the instance query vector; S343, input the instance query vector into the classification header to obtain the building damage classification result.

[0056] The following is a brief introduction to the training process of the building localization feature extraction network and the damage perception discrimination network: (1) A building damage assessment dataset xBD was constructed using publicly available disaster remote sensing image data, with image samples uniformly cropped to 512×512 pixels. This dataset contains pre-disaster and post-disaster image pairs, where pre-disaster images are used to provide prior information about building structures, and post-disaster images are used for building damage assessment. Data augmentation processing was performed on the post-disaster images, including random cropping, scaling, brightness perturbation, and rotation operations, to improve the model's generalization ability in complex disaster scenarios. Based on the building damage situation, the building damage levels were divided into four categories: undamaged, slightly damaged, moderately damaged, and severely damaged.

[0057] (2) Input the pre-disaster images into the building location feature extraction network to obtain the building location results, and input the post-disaster images and building location features into the damage perception and discrimination network to obtain the building damage classification results.

[0058] (3) Construct a first loss function based on the building location results, train the building location feature extraction network through the first loss function, construct a second loss function based on the building damage classification results, and train the damage perception and discrimination network through the second loss function.

[0059] The first loss function is:

[0060] in, These are segmentation loss, masking loss, and DICE loss. These are the weights corresponding to each loss; The second loss function is:

[0061] in, It is the actual label distribution. It is the probability distribution predicted for each category. These are the loss weights for each category. It is a true category index.

[0062] Based on the method described in the above embodiments, this embodiment will further describe the system from the perspective of a remote sensing image building disaster damage assessment system based on a visual basic model. This system can be implemented as an independent entity or integrated into an electronic device. The electronic device can be a terminal, server, or other devices. The terminal can include a tablet computer, a laptop computer, a personal computer (PC), a microprocessor box, or other devices.

[0063] Please see Figure 3 , Figure 3 This invention specifically describes a remote sensing image-based building disaster damage assessment system provided by an embodiment of the invention, which is applied to electronic devices. This system may include: The acquisition module is used to acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The positioning module is used to input the pre-disaster remote sensing image into a trained building positioning feature extraction network to obtain building positioning results. The building positioning feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building positioning feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building positioning features; inputting the building positioning features into the decoder to obtain positioning decoding features; and inputting the positioning decoding features into the mask prediction head to obtain a building mask. The classification module is used to input the post-disaster remote sensing images and the building location features into the trained damage perception and discrimination network to obtain the building damage classification results.

[0064] In specific implementation, the above modules and / or units can be implemented as independent entities, or they can be arbitrarily combined and implemented as the same or several entities. For the specific implementation of the above modules and / or units, please refer to the previous method embodiments. For the specific beneficial effects that can be achieved, please also refer to the beneficial effects in the previous method embodiments, which will not be repeated here.

[0065] In addition, this embodiment of the invention also provides an electronic device, which may be a computer, tablet computer, or other similar device. This electronic device can implement the steps in any embodiment of the visual-based remote sensing image building disaster damage assessment method provided by this invention. Therefore, it can achieve the beneficial effects that any visual-based remote sensing image building disaster damage assessment method provided by this invention can achieve, as detailed in the preceding embodiments, and will not be repeated here.

[0066] Figure 4 A specific structural block diagram of an electronic device provided in an embodiment of the present invention is shown. This electronic device can be used to implement the remote sensing image-based building disaster damage assessment method provided in the above embodiments. The electronic device 500 can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0067] The memory 520 can be used to store software programs and modules, such as the program instructions / modules corresponding to those in the above embodiments. The processor 580 executes various functional applications and data processing by running the software programs and modules stored in the memory 520. The memory 520 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 520 may further include memory remotely located relative to the processor 580, and these remote memories can be connected to the electronic device 500 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0068] The input unit 530 can be used to receive input numeric or character information, and to generate a keyboard and mouse related to user settings and function control. Display unit 540 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof. Display unit 540 may include display panel 541, which may optionally be configured in the form of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), or other similar forms.

[0069] Electronic device 500, through transmission module 570 (e.g., Wi-Fi module), can help users receive requests, send information, etc., providing users with wireless broadband internet access. Although transmission module 570 is shown in the figure, it is understood that it is not an essential component of electronic device 500 and can be omitted as needed without changing the essence of the invention.

[0070] The processor 580 is the control center of the electronic device 500. It connects to various parts of the phone via various interfaces and lines, and performs various functions and processes data of the electronic device 500 by running or executing software programs and / or modules stored in the memory 520, and by calling data stored in the memory 520, thereby providing overall monitoring of the electronic device. Optionally, the processor 580 may include one or more processing cores; in some embodiments, the processor 580 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 580.

[0071] Electronic device 500 also includes a power supply 590 (such as a battery) that supplies power to various components. In some embodiments, the power supply may be logically connected to processor 580 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 590 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0072] Although not shown, the electronic device 500 also includes cameras (such as front-facing cameras and rear-facing cameras), Bluetooth modules, etc., which will not be described in detail here. Specifically, in this embodiment, the display unit of the electronic device is a touch screen display, and the mobile terminal also includes a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors. One or more programs contain instructions for performing the following operations: Acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The pre-disaster remote sensing image is input into a trained building location feature extraction network to obtain building location results. The building location feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building location feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building location features, inputting the building location features into the decoder to obtain location decoding features, and inputting the location decoding features into the mask prediction head to obtain a building mask. The post-disaster remote sensing images and the building location features are input into a trained damage perception and discrimination network to obtain building damage classification results.

[0073] In practice, the above modules can be implemented as independent entities or combined in any way to be implemented as the same or several entities. For the specific implementation of the above modules, please refer to the previous method implementation examples, which will not be repeated here.

[0074] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. Therefore, embodiments of the present invention provide a storage medium storing multiple instructions that can be loaded by a processor to execute the steps of any embodiment of the remote sensing image-based building disaster damage assessment method provided by the present invention.

[0075] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0076] Since the instructions stored in the storage medium can execute the steps in any embodiment of the remote sensing image building disaster damage assessment method based on visual basic model provided in the embodiments of the present invention, the beneficial effects that any remote sensing image building disaster damage assessment method based on visual basic model provided in the embodiments of the present invention can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0077] The foregoing has provided a detailed description of a method, system, storage medium, and electronic device for assessing building disaster damage based on a visual fundamental model, as provided in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for building disaster damage assessment based on visual base model of remote sensing image, characterized in that, The method includes: Acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The pre-disaster remote sensing image is input into a trained building location feature extraction network to obtain building location results. The building location feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building location feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building location features, inputting the building location features into the decoder to obtain location decoding features, and inputting the location decoding features into the mask prediction head to obtain a building mask. The post-disaster remote sensing images and the building location features are input into a trained damage perception and discrimination network to obtain building damage classification results. 2.The method of claim 1, wherein, The encoder module includes a ViT image encoder and a CNN image encoder. The pre-disaster remote sensing image is input into the encoder module to obtain building location features, including: The pre-disaster remote sensing images are input into the ViT image encoder to obtain global features; The pre-disaster remote sensing images are input into the CNN image encoder to obtain local texture features; Based on the gating mechanism, the global features and the local texture features are fused to obtain building location features. 3.The method of claim 1, wherein, After obtaining the building mask, the process also includes: The building location result is obtained based on the spatial position corresponding to the building mask. 4.The method of claim 3, wherein, The pre-disaster remote sensing image is input into the ViT image encoder to obtain global features, which are expressed by the following formula: wherein, is a pre-disaster remote sensing image, is a ViT image encoder, is a global feature; The pre-disaster remote sensing images are input into the CNN image encoder to obtain local texture features, which are expressed by the following formula: wherein, is a CNN image encoder, is a local texture feature. 5.The vision-based model-based building damage assessment method using remote sensing images according to claim 1, wherein, The damage perception and discrimination network includes a ViT image encoder, a CNN image encoder, a decoder, and a classification head; The processing steps of the damage perception and discrimination network include: The post-disaster remote sensing images are input into the ViT image encoder to obtain texture change features, and the pre-disaster remote sensing images are input into the CNN image encoder to obtain context features. Based on the gating mechanism, the texture change features and the context features are fused to obtain the disaster change features; The damage characteristics are obtained by fusing the disaster change characteristics and the building location characteristics based on the channel attention mechanism and the spatial attention mechanism; The damage features are input into the decoder to obtain the instance query vector, and the instance query vector is input into the classification head to obtain the building damage classification result. 6.The method of claim 5, wherein the method further comprises: The damage features are input into the decoder to obtain an instance query vector, and the instance query vector is input into the classification head to obtain the building damage classification result, including: A set of learnable query vectors is predefined, with each vector representing a potential building instance; The learnable query vector and the damaged features are cross-attention calculated to obtain the instance query vector; The instance query vector is input into the classification header to obtain the building damage classification result. 7.The visual-based model building disaster damage assessment method based on remote sensing image according to claim 1, wherein, The method further includes: The building location feature extraction network is trained using a first loss function, which is: wherein, are respectively segmentation loss, mask loss and DICE loss, are respectively weights corresponding to each loss; The damage perception and discrimination network is trained using a second loss function, which is: where, is the true label distribution, is the predicted probability distribution for each class, is the loss weight for each class category, is the true class index. 8.A system for building disaster damage assessment based on visual base model of remote sensing image, characterized in that, include: The acquisition module is used to acquire pre-disaster and post-disaster remote sensing images of the building disaster area; The positioning module is used to input the pre-disaster remote sensing image into a trained building positioning feature extraction network to obtain building positioning results. The building positioning feature extraction network includes an encoder module, a decoder, and a mask prediction head. The processing steps of the building positioning feature extraction network include: inputting the pre-disaster remote sensing image into the encoder module to obtain building positioning features; inputting the building positioning features into the decoder to obtain positioning decoding features; and inputting the positioning decoding features into the mask prediction head to obtain a building mask. The classification module is used to input the post-disaster remote sensing images and the building location features into the trained damage perception and discrimination network to obtain the building damage classification results.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted to be loaded by a processor to execute the remote sensing image building disaster damage assessment method based on a visual fundamental model as described in any one of claims 1 to 7.

10. An electronic device, comprising: The device includes a processor and a memory, the processor being electrically connected to the memory, the memory being used to store instructions and data, and the processor being used to execute the steps in the method for assessing building disaster damage based on a visual fundamental model according to any one of claims 1 to 7.

Images