elliotvincent/b-FLAIR

b-FLAIR: bi-temporal extension of FLAIR

<img src="./thumbnail.png" alt="b-FLAIR" width="700">

Dataset Description

b-FLAIR is a temporal extension of the FLAIR dataset [1] focused on land cover classification in France. The dataset provides bi-temporal orthoimage pairs with single-temporal semantic annotations.

Project page: https://xavibou.github.io/CDviaWTS/

Dataset Summary

• —Task: Semantic change detection via weak temporal supervision
• —Coverage: France
• —Resolution: 0.2 m/px
• —Patch Size: 512×512 pixels
• —Bands: Red, Green, Blue, Infrared, Elevation
• —Total Training Pairs: 61,712
• —Total Test Pairs: 16,050
• —Validation split: Corresponds to folders D004, D014, D029, D031, D058, D066, D067, D077 in train

Dataset Creation

Source Data

The dataset extends the original FLAIR dataset [1] by adding new orthoimage acquisitions over the same geographic locations. New images were downloaded from IGN's BD ORTHO [2]. 82.9% of our added images correspond to images acquired after the corresponding original FLAIR patches, while the remaining were acquired before.

Annotations

Original FLAIR single-temporal semantic masks are provided for each pair. They classify each pixel of t1 images in one of 19 semantic land cover classes. Please refer to [1] for more information on these annotations.

### [!IMPORTANT] Important In order to avoid duplicates, this repo does not contain original FLAIR images and labels. Data for imgt1 and labelst1 should be downloaded separately from the links below: - imgt1/train: [Download link](https://storage.gra.cloud.ovh.net/v1/AUTH366279ce616242ebb14161b7991a8461/defi-ia/flairdata2/flairaerialtrain.zip) - imgt1/test: [Download link](https://storage.gra.cloud.ovh.net/v1/AUTH366279ce616242ebb14161b7991a8461/defi-ia/flairdata2/flair2aerialtest.zip) - labelst1/train: Donwload link - labelst1/test: [Download link](https://storage.gra.cloud.ovh.net/v1/AUTH366279ce616242ebb14161b7991a8461/defi-ia/flairdata2/flair2labels_test.zip)

References

[1] Garioud et al. (2023). FLAIR: a country-scale land cover semantic segmentation dataset from multi-source optical imagery. In NeurIPS [2] IGN - Institut national de l’information géographique et forestière. (2025). BD ORTHO®: L’image géographique du territoire national, la France vue du ciel.

Citation

If you use this dataset, please cite the following publication:

@article{bou2026remote,
  title={Remote Sensing Change Detection via Weak Temporal Supervision},
  author={Bou, Xavier and Vincent, Elliot and Facciolo, Gabriele and Grompone von Gioi, Rafael and Morel, Jean-Michel and Ehret, Thibaud},
  journal={arXiv preprint arXiv:2601.02126},
  year={2026}
}