README.md

metadata

license: etalab-2.0
pipeline_tag: image-segmentation
library_name: pytorch
tags:
  - semantic segmentation
  - pytorch
  - landcover
model-index:
  - name: FLAIR-HUB_LPIS-F_utae
    results:
      - task:
          type: semantic-segmentation
        dataset:
          name: IGNF/FLAIR-HUB/
          type: earth-observation-dataset
        metrics:
          - type: mIoU
            value: 21.755
            name: mIoU
          - type: OA
            value: 85.282
            name: Overall Accuracy
          - type: IoU
            value: 83.86
            name: IoU building
          - type: IoU
            value: 78.38
            name: IoU greenhouse
          - type: IoU
            value: 61.59
            name: IoU swimming pool
          - type: IoU
            value: 61.59
            name: IoU impervious surface
          - type: IoU
            value: 57.17
            name: IoU pervious surface
          - type: IoU
            value: 62.94
            name: IoU bare soil
          - type: IoU
            value: 90.35
            name: IoU water
          - type: IoU
            value: 63.38
            name: IoU snow
          - type: IoU
            value: 54.34
            name: IoU herbaceous vegetation
          - type: IoU
            value: 57.14
            name: IoU agricultural land
          - type: IoU
            value: 34.85
            name: IoU plowed land
          - type: IoU
            value: 24.517
            name: IoU vineyard
          - type: IoU
            value: 71.73
            name: IoU deciduous
          - type: IoU
            value: 62.6
            name: IoU coniferous
          - type: IoU
            value: 30.19
            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.	🛩️ Historical
LC-A	✓		✓
LC-D	✓		✓			✓
LC-F	✓		✓			✓	✓
LC-G	✓					✓
LC-I	✓				✓
LC-L	✓		✓	✓	✓	✓	✓
LPIS-A		✓	✓
LPIS-F		✓				✓
LPIS-I		✓			✓	✓	✓
LPIS-J		✓	✓		✓	✓	✓

🔍 Model: FLAIR-HUB_LPIS-F_utae

• Encoder: UTAE
• Decoder: UTAE
• Metrics:

mIoU	O.A.	F-score	Precision	Recall
21.75%	85.28%	28.74%	28.98%	31.90%

• Params.: 0.9

---

General Informations

• Contact: flair@ign.fr
• Code repository: https://github.com/IGNF/FLAIR-HUB
• Paper: https://arxiv.org/abs/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: UTAE
- 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:
    - SENTINEL2_TS : [1,2,3,4,5,6,7,8,9,10]

---

Training Data

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

---

Training Logging

---

Metrics

Metric	Value
mIoU	21.75%
Overall Accuracy	85.28%
F-score	28.74%
Precision	28.98%
Recall	31.90%

Class	IoU (%)	F-score (%)	Precision (%)	Recall (%)
grasses	43.13	60.27	64.86	56.28
wheat	59.61	74.70	66.33	85.47
barley	48.82	65.61	73.01	59.57
maize	68.82	81.53	73.99	90.78
other cereals	2.60	5.08	15.86	3.02
rice	0.00	0.00	0.00	0.00
flax/hemp/tobacco	0.00	0.00	0.00	0.00
sunflower	27.98	43.73	48.70	39.68
rapeseed	70.93	82.99	76.64	90.49
other oilseed crops	0.00	0.00	0.00	0.00
soy	12.37	22.02	14.49	45.84
other protein crops	20.86	34.52	27.86	45.35
fodder legumes	22.85	37.20	28.70	52.83
beetroots	1.51	2.98	17.46	1.63
potatoes	0.00	0.00	0.00	0.00
other arable crops	10.06	18.28	13.58	27.97
vineyard	24.52	39.38	37.92	40.96
olive groves	0.00	0.00	0.00	0.00
fruits orchards	0.00	0.00	0.00	0.00
nut orchards	0.00	0.00	0.00	0.00
other permanent crops	0.00	0.00	0.00	0.00
mixed crops	0.03	0.05	15.72	0.03
background	86.27	92.63	91.41	93.88

---

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