Scholarlink

Land cover categorization offers current data on the Earth’s resources, which is crucial for agricultural research, urban planning, and disaster surveillance. The spatial-spectral, radiometric, and temporal resolutions of pictures have been enhanced throughout time due to the recent advancements in sensor technology on satellite and aerial remote sensing (RS) systems. These enhancements provide great opportunities for comprehending land cover data. Moreover, it is necessary to make use of all accessible satellite pictures to identify various land cover categories and track their changes over time, regardless of their spatial, spectral, temporal, and radiometric resolutions. Various deep learning models demonstrate high efficiency and accuracy in the classification of satellite pictures depicting land cover. Therefore, multiple models including VGG-16, ResNet-50, and YOLOv8 were utilized for the comparison study. Simulation research demonstrates that YOLOv8 has attained the greatest accuracy of 98% among the three models.

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

Land Cover Classification: A Comparative Analysis for Deep Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rohan Vaghela,
  • Jigar Sarda,
  • Amit Thakkar,
  • Biswajit Brahma,
  • Akash Kumar Bhoi,
  • Hirva Patel,
  • Arpita Patel

摘要

Land cover categorization offers current data on the Earth’s resources, which is crucial for agricultural research, urban planning, and disaster surveillance. The spatial-spectral, radiometric, and temporal resolutions of pictures have been enhanced throughout time due to the recent advancements in sensor technology on satellite and aerial remote sensing (RS) systems. These enhancements provide great opportunities for comprehending land cover data. Moreover, it is necessary to make use of all accessible satellite pictures to identify various land cover categories and track their changes over time, regardless of their spatial, spectral, temporal, and radiometric resolutions. Various deep learning models demonstrate high efficiency and accuracy in the classification of satellite pictures depicting land cover. Therefore, multiple models including VGG-16, ResNet-50, and YOLOv8 were utilized for the comparison study. Simulation research demonstrates that YOLOv8 has attained the greatest accuracy of 98% among the three models.