Scholarlink

Detecting and identifying gully erosion is crucial for effective erosion risk management and dam siltation planning for monitoring worldwide. In this research, we used modern deep-learning method to detect gully erosion in the western rif part of morocco. we applied an advanced model, namely Segformer. Therefore, high resolution of satellite imagery from 2023 was used, and a gully erosion mask was created using ArcGIS Pro. The model was based on generated data, which was cropped into small patches of 300 images and 300 masks with a size of 256 × 256 pixels. We split the data into 80% for training and 20% for validation. The results showed that this deep-learning model perform well in detecting gully erosion. Our model achieved an accuracy of 97% and a low loss value of 0.0293, showing that it performed well during evaluation. This research provides valuable tools for land management, soil conservation, and dam siltation prevention by accurately identifying areas of gully erosion. This model is proving to be more effective than traditional methods with its ability to capture complex patterns in satellite imagery.

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

Deep Learning Semantic Segmentation for Gully Erosion Detection in Northern Morocco Using Segformer Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Manar El Kihel,
  • Abdessalam Ouallali,
  • Mohamed Rabii Simou,
  • Soufiane Aafir,
  • Laila Saafadi

摘要

Detecting and identifying gully erosion is crucial for effective erosion risk management and dam siltation planning for monitoring worldwide. In this research, we used modern deep-learning method to detect gully erosion in the western rif part of morocco. we applied an advanced model, namely Segformer. Therefore, high resolution of satellite imagery from 2023 was used, and a gully erosion mask was created using ArcGIS Pro. The model was based on generated data, which was cropped into small patches of 300 images and 300 masks with a size of 256 × 256 pixels. We split the data into 80% for training and 20% for validation. The results showed that this deep-learning model perform well in detecting gully erosion. Our model achieved an accuracy of 97% and a low loss value of 0.0293, showing that it performed well during evaluation. This research provides valuable tools for land management, soil conservation, and dam siltation prevention by accurately identifying areas of gully erosion. This model is proving to be more effective than traditional methods with its ability to capture complex patterns in satellite imagery.