Zhejiang University (20240212162). SEMANTIC SEGMENTATION METHOD FOR CROSS-SATELLITE REMOTE SENSING IMAGES BASED ON UNSUPERVISED BIDIRECTIONAL DOMAIN ADAPTATION AND FUSION simplified abstract

From WikiPatents
Jump to navigation Jump to search

SEMANTIC SEGMENTATION METHOD FOR CROSS-SATELLITE REMOTE SENSING IMAGES BASED ON UNSUPERVISED BIDIRECTIONAL DOMAIN ADAPTATION AND FUSION

Organization Name

Zhejiang University

Inventor(s)

JIANWEI Yin of HANGZHOU, ZHEJIANG PROVINCE (CN)

YUXIANG Cai of HANGZHOU, ZHEJIANG PROVINCE (CN)

YINGCHUN Yang of HANGZHOU, ZHEJIANG PROVINCE (CN)

SHUIGUANG Deng of HANGZHOU, ZHEJIANG PROVINCE (CN)

YING Li of HANGZHOU, ZHEJIANG PROVINCE (CN)

SEMANTIC SEGMENTATION METHOD FOR CROSS-SATELLITE REMOTE SENSING IMAGES BASED ON UNSUPERVISED BIDIRECTIONAL DOMAIN ADAPTATION AND FUSION - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240212162 titled 'SEMANTIC SEGMENTATION METHOD FOR CROSS-SATELLITE REMOTE SENSING IMAGES BASED ON UNSUPERVISED BIDIRECTIONAL DOMAIN ADAPTATION AND FUSION

The present invention introduces a method for semantic segmentation of cross-satellite remote sensing images using unsupervised bidirectional domain adaptation and fusion.

  • Training of bidirectional source-target domain image translation models
  • Selection of bidirectional generators in the image translation models
  • Bidirectional translation of source-target domain images
  • Training of source and target domain semantic segmentation models
  • Generation and fusion of source and target domain segmentation probabilities

By employing source-target and target-source bidirectional domain adaptation, the method enhances the accuracy and robustness of semantic segmentation models for cross-satellite remote sensing images. Additionally, the bidirectional semantic consistency loss and generator parameter selection help mitigate instability issues in the image translation models.

      1. Potential Applications:

This technology can be applied in various fields such as environmental monitoring, urban planning, disaster management, and agriculture.

      1. Problems Solved:

1. Improved accuracy and robustness of semantic segmentation models for cross-satellite remote sensing images. 2. Mitigation of instability problems in bidirectional image translation models.

      1. Benefits:

1. Enhanced accuracy in semantic segmentation of remote sensing images. 2. Increased robustness in handling different satellite image datasets.

      1. Commercial Applications:

The technology can be utilized by satellite imaging companies, environmental agencies, urban planners, and agricultural organizations to improve their image analysis capabilities and decision-making processes.

      1. Questions about Semantic Segmentation of Cross-Satellite Remote Sensing Images:

1. How does bidirectional domain adaptation improve the accuracy of semantic segmentation models? 

  - Bidirectional domain adaptation helps in aligning the source and target domain features, leading to enhanced model performance.

2. What are the key benefits of using unsupervised bidirectional domain adaptation in remote sensing image analysis? 

  - Unsupervised bidirectional domain adaptation allows for the transfer of knowledge between different satellite datasets without the need for labeled data, improving the generalization of the models.


Original Abstract Submitted

the present invention discloses a semantic segmentation method for cross-satellite remote sensing images based on unsupervised bidirectional domain adaptation and fusion. the method includes training of bidirectional source-target domain image translation models, selection of bidirectional generators in the image translation models, bidirectional translation of source-target domain images, training of source and target domain semantic segmentation models, and generation and fusion of source and target domain segmentation probabilities. according to the present invention, by utilizing source-target and target-source bidirectional domain adaptation, the source and target domain segmentation probabilities are fused, which improves the accuracy and robustness of a semantic segmentation model for the cross-satellite remote sensing images; and further, through the bidirectional semantic consistency loss and the selection of the parameters of the generators, the influence due to the instability problem of the generators in the bidirectional image translation models is avoided.

Retrieved from "http://wikipatents.org/index.php?title=Zhejiang_University_(20240212162)._SEMANTIC_SEGMENTATION_METHOD_FOR_CROSS-SATELLITE_REMOTE_SENSING_IMAGES_BASED_ON_UNSUPERVISED_BIDIRECTIONAL_DOMAIN_ADAPTATION_AND_FUSION_simplified_abstract&oldid=82135"