AIDERv2 (Aerial Image Dataset for Emergency Response)
- Source
- Zenodo
- Published
- 2024
- Format
- multiple formats
- Access
- Open access
- License
- CC-BY-4.0
- Themes
- Natural Disasters, Machine Learning & AI, Remote Sensing & Satellites
About this dataset
This dataset consists of 167,723 aerial images divided into four classes: earthquake/collapsed buildings, flood, wildfire/fire, and a normal class. It is designed for training convolutional neural networks to detect natural disasters from aerial images.
How the data was collected
Aerial images collected and labeled to represent different disaster scenarios, facilitating machine learning applications in emergency response. The dataset includes metadata for each image to assist in training and evaluation processes.