Scholarlink

<p>As a slope hazard that seriously affects mountainous people, landslides always have inheritance and repetition. Due to the rapid change in land use/cover and the poor performance of remote sensing technologies, the landslide-monitoring task becomes more difficult. Ongoing monitoring endeavors have the potential to enhance the precision and effectiveness of models. Therefore, this study looks at several semantic segmentation models and their backbone architecture and input size in order to find landslides using Worldview-2 images. An indicator system to detect landslide traces on the field and very high-resolution images was proposed in detail. An examination of the training and assessment of U-Net, DeepLab-v3, and PSPN indicates that DeepLab-v3 exhibits consistently high accuracy (from 96 to 99%). As a result, DeepLab-v3 models with a DenseNet backbone can identify landslides at diverse input sizes, especially with the size of 256 × 256. These architectures are incorporated into deep-learning models to improve the frequency and spatial resolution of landslide detection. It aids in disaster prevention and management through real-time monitoring and updating of landslide risk maps aids in disaster prevention and management. The integration of trained DeepLab-v3 models and 0.7&#xa0;m-resolution remote sensing data enhanced real-time landslide risk map monitoring and updates in the Northwestern provinces of Vietnam. As a result, the trained model helps identify unreported landslides from WorldView-2 imagery in remote locations, away from both roads and residential places, which stand out of reach during field surveys. The outcomes support policymakers and researchers to develop better landslide risk mitigation approaches and response strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploring Various Deep Learning Models for High-Precision Landslide Tracing in Very-High Resolution Remote Sensing Imagery

  • 期刊论文

  • 发表日期：

    2025年7月10日

查看全文
  • Hieu Nguyen,
  • Kinh Bac Dang,
  • Tuan Linh Giang,
  • Van Bao Dang,
  • Nguyen Vu Dang,
  • Minh Duc Dao

摘要

As a slope hazard that seriously affects mountainous people, landslides always have inheritance and repetition. Due to the rapid change in land use/cover and the poor performance of remote sensing technologies, the landslide-monitoring task becomes more difficult. Ongoing monitoring endeavors have the potential to enhance the precision and effectiveness of models. Therefore, this study looks at several semantic segmentation models and their backbone architecture and input size in order to find landslides using Worldview-2 images. An indicator system to detect landslide traces on the field and very high-resolution images was proposed in detail. An examination of the training and assessment of U-Net, DeepLab-v3, and PSPN indicates that DeepLab-v3 exhibits consistently high accuracy (from 96 to 99%). As a result, DeepLab-v3 models with a DenseNet backbone can identify landslides at diverse input sizes, especially with the size of 256 × 256. These architectures are incorporated into deep-learning models to improve the frequency and spatial resolution of landslide detection. It aids in disaster prevention and management through real-time monitoring and updating of landslide risk maps aids in disaster prevention and management. The integration of trained DeepLab-v3 models and 0.7 m-resolution remote sensing data enhanced real-time landslide risk map monitoring and updates in the Northwestern provinces of Vietnam. As a result, the trained model helps identify unreported landslides from WorldView-2 imagery in remote locations, away from both roads and residential places, which stand out of reach during field surveys. The outcomes support policymakers and researchers to develop better landslide risk mitigation approaches and response strategies.