RescueNet: A High Resolution UAV Semantic Segmentation Benchmark Dataset for Natural Disaster Damage Assessment

Image | Hurricane | Response

RescueNet is a dataset for hurricane damage assessmnet. It contains 4,494 post disaster images collected after Hurricane Michael and is introduced for semantic segmentation.

  • ML task type: Semantic segmentation
  • Data Source: Earth Observation Data and GeoSpatial Imagery (UAV)
  • Size: 4,494 Images
  • Timespan: October 11-14, 2018
  • Geographical Coverage: Mexico Beach and other areas directly affected by Hurricane Michael
  • Baseline Information
  • Evaluated on: Attention U-Net
  • Metrics used: Mean Intersection over Union
  • Results as reported in original paper: mIOU: 93.98%

Tashnim Chowdhury, Robin Murphy, and Maryam Rahnemoonfar. RescueNet: A high resolution uav semantic segmentation benchmark dataset for natural disaster damage assessment. arXiv preprint arXiv:2202.12361, 2022