AIDERv2

Overview

AIDERv2 is an aerial image classification dataset for recognizing natural disasters and non-disaster scenes. It contains 16,723 images across four classes: earthquake, flood, normal, and wildfire. The dataset extends the original AIDER dataset with additional open-source imagery, including frames extracted from YouTube videos.

Train: 13,399 images
Validation: 1,670 images
Test: 1,654 images

Data Structure

When accessing an example using ds[i], you will receive a dictionary with the following keys:

Key	Type	Description
`image`	`PIL.Image.Image`	Aerial image at its original dimensions
`label`	int	Class label (0-3), corresponding to the class mapping below

Class Labels

Label	Class	Description
0	Earthquake	Images depicting earthquake-related damage
1	Flood	Images depicting flooding
2	Normal	Images without a depicted disaster
3	Wildfire	Images depicting wildfire or fire

Usage Example

Basic Usage

from stable_datasets.images.aiderv2 import AIDERv2

# Load individual splits
train_ds = AIDERv2(split="train")
val_ds = AIDERv2(split="validation")
test_ds = AIDERv2(split="test")

sample = train_ds[0]
print(sample.keys())  # {"image", "label"}

# Inspect class names
print(train_ds.info.features["label"].names)

# Optional: make it PyTorch-friendly
train_torch = train_ds.with_format("torch")

References

Dataset: https://zenodo.org/records/10891054
License: Creative Commons Attribution 4.0 International

Citation

@dataset{shianios_kyrkou_2024_aiderv2,
  author = {Shianios, D. and Kyrkou, C.},
  title = {AIDERv2 (Aerial Image Dataset for Emergency Response Applications)},
  year = {2024},
  publisher = {Zenodo},
  doi = {10.5281/zenodo.10891054},
  url = {https://doi.org/10.5281/zenodo.10891054}
}

 @inproceedings{shianios2023benchmark,
   title={A Benchmark and Investigation of Deep-Learning-Based Techniques for Detecting Natural Disasters in Aerial Images},
   author={Shianios, D. and Kyrkou, C. and Kolios, P. S.},
   booktitle={Computer Analysis of Images and Patterns (CAIP)},
   series={Lecture Notes in Computer Science},
   volume={14185},
   publisher={Springer},
   year={2023},
   doi={10.1007/978-3-031-44240-7_24}
 }

 @article{shianios2024direcnetv2,
   title={DiRecNetV2: A Transformer-Enhanced Network for Aerial Disaster Recognition},
   author={Shianios, D. and Kolios, P. and Kyrkou, C.},
   journal={SN Computer Science},
   year={2024}
 }