Scholarlink

Increasing frequency and severity of floods demands accurate prediction and damage evaluation to effective disaster response. Traditional systems often lack accurate information on time, causing delay, and disable action. This research implements intensive learning, especially CNN architecture, and analyses satellite imagery to detect floods. Transfer learning increases accuracy by reducing computational costs and obtains classification accuracy of 97%. Despite the success, challenges remain, including limited access to high-resolution images and difficulty in flood types and generalization in regions. High computational demands also obstruct real-time use. Future work models will focus on normalization, efficiency, and improving data access. Overall, deep learning shows reliable, rapid flood prediction, and strong potential for damage evaluation, increasing disaster preparations and reaction.

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-Based Disaster Prediction and Damage Assessment Using Satellite Imagery and Transfer Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • S. Selva Kumara Samy,
  • Vallabh Sumanth,
  • Shashank Rajput,
  • N. Gopinath,
  • C. G. Anupama,
  • C. Arun

摘要

Increasing frequency and severity of floods demands accurate prediction and damage evaluation to effective disaster response. Traditional systems often lack accurate information on time, causing delay, and disable action. This research implements intensive learning, especially CNN architecture, and analyses satellite imagery to detect floods. Transfer learning increases accuracy by reducing computational costs and obtains classification accuracy of 97%. Despite the success, challenges remain, including limited access to high-resolution images and difficulty in flood types and generalization in regions. High computational demands also obstruct real-time use. Future work models will focus on normalization, efficiency, and improving data access. Overall, deep learning shows reliable, rapid flood prediction, and strong potential for damage evaluation, increasing disaster preparations and reaction.