metadata

license: etalab-2.0
pipeline_tag: image-segmentation
library_name: pytorch
tags:
  - semantic segmentation
  - pytorch
  - landcover
model-index:
  - name: FLAIR-HUB_LC-A_RVB_swinbase-upernet
    results:
      - task:
          type: semantic-segmentation
        dataset:
          name: IGNF/FLAIR-HUB/
          type: earth-observation-dataset
        metrics:
          - type: mIoU
            value: 64.054
            name: mIoU
          - type: OA
            value: 76.784
            name: Overall Accuracy
          - type: IoU
            value: 83.769
            name: IoU building
          - type: IoU
            value: 77.891
            name: IoU greenhouse
          - type: IoU
            value: 61.59
            name: IoU swimming pool
          - type: IoU
            value: 75.029
            name: IoU impervious surface
          - type: IoU
            value: 56.972
            name: IoU pervious surface
          - type: IoU
            value: 65.214
            name: IoU bare soil
          - type: IoU
            value: 90.08
            name: IoU water
          - type: IoU
            value: 67.767
            name: IoU snow
          - type: IoU
            value: 52.851
            name: IoU herbaceous vegetation
          - type: IoU
            value: 56.529
            name: IoU agricultural land
          - type: IoU
            value: 37.34
            name: IoU plowed land
          - type: IoU
            value: 78.876
            name: IoU vineyard
          - type: IoU
            value: 70.071
            name: IoU deciduous
          - type: IoU
            value: 58.948
            name: IoU coniferous
          - type: IoU
            value: 30.973
            name: IoU brushwood

🌐 FLAIR-HUB Model Collection

Trained on: FLAIR-HUB dataset 🔗
Available modalities: Aerial images, SPOT images, Topographic info, Sentinel-2 yearly time-series, Sentinel-1 yearly time-series, Historical aerial images
Encoders: ConvNeXTV2, Swin (Tiny, Small, Base, Large)
Decoders: UNet, UPerNet
Tasks: Land-cover mapping (LC), Crop-type mapping (LPIS)
Class nomenclature: 15 classes for LC, 23 classes for LPIS

🆔 Model ID	🗺️ Land-cover	🌾 Crop-types	🛩️ Aerial	⛰️ Elevation	🛰️ SPOT	🛰️ S2 t.s.	🛰️ S1 t.s.
LC-A	✓		✓
LC-D	✓		✓			✓
LC-F	✓		✓			✓	✓
LC-G	✓					✓
LC-I	✓				✓
LC-L	✓		✓	✓	✓	✓	✓
LPIS-A		✓	✓
LPIS-F		✓				✓
LPIS-I		✓			✓	✓	✓
LPIS-J		✓	✓		✓	✓	✓

🔍 Model: FLAIR-HUB_LC-A_RVB_swinbase-upernet

Encoder: swin_base_patch4_window12_384
Decoder: upernet
Metrics:

mIoU	O.A.	F-score	Precision	Recall
64.05%	76.78%	76.88%	77.71%	76.59%

Params.: 89.4

General Informations

Contact: [email protected]
Code repository: https://github.com/IGNF/FLAIR-HUB
Paper: https://huggingface.co/papers/2506.07080
Developed by: IGN
Compute infrastructure:
- software: python, pytorch-lightning
- hardware: HPC/AI resources provided by GENCI-IDRIS
License: Etalab 2.0

Training Config Hyperparameters

- Model architecture: swin_base_patch4_window12_384-upernet
- Optimizer: AdamW (betas=[0.9, 0.999], weight_decay=0.01)
- Learning rate: 5e-5
- Scheduler: one_cycle_lr (warmup_fraction=0.2)
- Epochs: 150
- Batch size: 5
- Seed: 2025
- Early stopping: patience 20, monitor val_miou (mode=max)
- Class weights:
    - default: 1.0
    - masked classes: [clear cut, ligneous, mixed, other] → weight = 0
- Input channels:
    - AERIAL_RGBI : [1,2,3]
- Input normalization (custom):
    - AERIAL_RGBI:
        mean: [105.66, 111.35, 102.18 ]
        std:  [52.23, 45.62, 44.30]

Training Data

- Train patches: 152225
- Validation patches: 38175
- Test patches: 50700

Training Logging

Metrics

Metric	Value
mIoU	64.05%
Overall Accuracy	76.78%
F-score	76.88%
Precision	77.71%
Recall	76.59%

Class	IoU (%)	F-score (%)	Precision (%)	Recall (%)
building	83.77	91.17	91.42	90.92
greenhouse	77.89	87.57	85.28	89.99
swimming pool	58.50	73.82	77.36	70.58
impervious surface	75.03	85.73	87.13	84.38
pervious surface	56.97	72.59	70.18	75.17
bare soil	65.21	78.94	74.64	83.78
water	90.08	94.78	95.00	94.57
snow	67.77	80.79	97.53	68.95
herbaceous vegetation	52.85	69.15	71.87	66.64
agricultural land	56.53	72.23	68.13	76.85
plowed land	37.34	54.38	51.25	57.91
vineyard	78.88	88.19	86.89	89.53
deciduous	70.07	82.40	81.00	83.85
coniferous	58.95	74.17	79.78	69.30
brushwood	30.97	47.30	48.20	46.43

Inference

Aerial ROI

Inference ROI

Cite

BibTeX:

@article{ign2025flairhub,
  doi = {10.48550/arXiv.2506.07080},
  url = {https://arxiv.org/abs/2506.07080},
  author = {Garioud, Anatol and Giordano, Sébastien and David, Nicolas and Gonthier, Nicolas},
  title = {FLAIR-HUB: Large-scale Multimodal Dataset for Land Cover and Crop Mapping},
  publisher = {arXiv},
  year = {2025}
}

APA:

Anatol Garioud, Sébastien Giordano, Nicolas David, Nicolas Gonthier. 
FLAIR-HUB: Large-scale Multimodal Dataset for Land Cover and Crop Mapping. (2025). 
DOI: https://doi.org/10.48550/arXiv.2506.07080