IGNF/FLAIR-HUB_LC-L_swinbase-upernet

🌐 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.	🛩️ Historical
LC-A	✓		✓
LC-D	✓		✓			✓
LC-F	✓		✓			✓	✓
LC-G	✓					✓
LC-I	✓				✓
LC-L	✓		✓	✓	✓	✓	✓
LPIS-A		✓	✓
LPIS-F		✓				✓
LPIS-I		✓			✓	✓	✓
LPIS-J		✓	✓		✓	✓	✓

🔍 Model: FLAIR-HUB_LC-L_swinbase-upernet

• Encoder: swin_base_patch4_window12_384
• Decoder: upernet
• Metrics:

mIoU	O.A.	F-score	Precision	Recall
65.78%	78.24%	78.15%	78.28%	78.30%

• Params.: 276.4

---

General Informations

• Contact: [email protected]
• Code repository: https://github.com/IGNF/FLAIR-HUB
• Paper: https://arxiv.org/abs/2506.07080
• Project Page: https://ignf.github.io/FLAIR/FLAIR-HUB/flairhub
• 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: [4, 1, 2]
    - SPOT_RGBI: [4, 1, 2]
    - SENTINEL2_TS: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
    - SENTINEL1-ASC_TS: [1, 2]
    - SENTINEL1-DESC_TS: [1, 2]
- Input normalization (custom):
    - AERIAL_RGBI:
        mean: [106.59, 105.66, 111.35]
        std:  [39.78, 52.23, 45.62]
    - SPOT_RGBI:
        mean: [1137.03, 433.26, 508.75]
        std:  [543.11, 312.76, 284.61]
    - DEM_ELEV:
        means: [311.06, 311.06]
        std: [537.55, 537.55]

---

Training Data

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

---

Training Logging

---

Metrics

Metric	Value
mIoU	65.78%
Overall Accuracy	78.24%
F-score	78.15%
Precision	78.28%
Recall	78.30%

Class	IoU (%)	F-score (%)	Precision (%)	Recall (%)
building	85.31	92.07	92.12	92.02
greenhouse	79.09	88.32	84.61	92.38
swimming pool	62.04	76.57	74.86	78.36
impervious surface	76.58	86.74	86.65	86.82
pervious surface	58.25	73.62	72.56	74.71
bare soil	64.66	78.54	75.42	81.92
water	90.52	95.02	95.87	94.20
snow	73.41	84.67	96.42	75.47
herbaceous vegetation	55.13	71.08	72.57	69.64
agricultural land	58.63	73.92	70.75	77.39
plowed land	37.55	54.60	53.74	55.48
vineyard	78.56	87.99	85.42	90.72
deciduous	72.26	83.90	83.69	84.10
coniferous	63.55	77.71	79.25	76.23
brushwood	31.12	47.47	50.24	44.99

---

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