Upload folder using huggingface_hub (part 2)

.gitattributes

@@ -5,3 +5,5 @@
 *.safetensors filter=lfs diff=lfs merge=lfs -text
 models/finetuned/landcover/satquery_landcover_v1/confusion_matrix.png filter=lfs diff=lfs merge=lfs -text
 models/finetuned/landcover/satquery_landcover_v1/qualitative_overlays.png filter=lfs diff=lfs merge=lfs -text
+models/land_cover/flair-rgb-15cl-deeplabv3/confmat.npy filter=lfs diff=lfs merge=lfs -text
+models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/model_utils/AerialROI.png filter=lfs diff=lfs merge=lfs -text

models/finetuned/water/satquery_water_shadow_v1/satquery_water_shadow_state_dict.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e93903a19a09b98d9d8c2ac3f9973491f03359c679e8fc5cda3c9c8432b956d5
+size 360362139

models/finetuned/water/satquery_water_shadow_v1/thresholds.json

@@ -0,0 +1,5 @@
+{
+  "water_probability_threshold": 0.6,
+  "minimum_component_area_pixels": 16,
+  "uncertainty_threshold": 0.3
+}

models/land_cover/flair-rgb-15cl-deeplabv3/.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

models/land_cover/flair-rgb-15cl-deeplabv3/FLAIR-INC_rgb_15cl_resnet34-deeplabv3_weights.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b872ab0a4940bf1344e194df6e9a71b1f1847d5771ae450459cabacd42a05541
+size 104221495

models/land_cover/flair-rgb-15cl-deeplabv3/README.md

@@ -0,0 +1,322 @@
+---
+license: etalab-2.0
+tags:
+- segmentation
+- pytorch
+- aerial imagery
+- landcover
+- IGN
+model-index:
+- name: FLAIR-INC_rgb_15cl_resnet34-deeplabv3
+  results:
+  - task:
+      type: semantic-segmentation
+    dataset:
+      name: IGNF/FLAIR/
+      type: earth-observation-dataset
+    metrics:
+    - name: mIoU
+      type: mIoU
+      value: 58.093
+    - name: Overall Accuracy
+      type: OA
+      value: 74.887
+    - name: Fscore
+      type: Fscore
+      value: 72.331
+    - name: Precision
+      type: Precision
+      value: 72.577
+    - name: Recall
+      type: Recall
+      value: 73.390
+      
+    - name: IoU Buildings
+      type: IoU
+      value: 74.671
+    - name: IoU Pervious surface
+      type: IoU
+      value: 52.890
+    - name: IoU Impervious surface
+      type: IoU
+      value: 69.452
+    - name: IoU Bare soil
+      type: IoU
+      value: 59.280
+    - name: IoU Water
+      type: IoU
+      value: 82.144
+    - name: IoU Coniferous
+      type: IoU
+      value: 55.443
+    - name: IoU Deciduous
+      type: IoU
+      value: 68.459 
+    - name: IoU Brushwood
+      type: IoU
+      value: 28.215
+    - name: IoU Vineyard
+      type: IoU
+      value: 77.997
+    - name: IoU Herbaceous vegetation
+      type: IoU
+      value: 50.624      
+    - name: IoU Agricultural land
+      type: IoU
+      value: 58.221
+    - name: IoU Plowed land
+      type: IoU
+      value: 43.881   
+    - name: IoU Swimming pool
+      type: IoU
+      value: 38.207    
+    - name: IoU Greenhouse
+      type: IoU
+      value: 53.815
+      
+pipeline_tag: image-segmentation
+---
+
+
+<div style="border:0px; padding:25px; background-color:#F8F5F5; padding-top:10px; padding-bottom:1px;">
+  <h1>FLAIR model collection</h1>
+  <p>The FLAIR models are a collection of semantic segmentation models initially developed to classify land cover on very high resolution aerial images (more specifically the French <a href="https://geoservices.ign.fr/bdortho">BD ORTHO®</a> product). The distributed pre-trained models differ in their :</p>
+  <ul style="list-style-type:disc;">
+    <li>dataset for training : <a href="https://huggingface.co/datasets/IGNF/FLAIR"><b>FLAIR</b> dataset</a> or the increased version of this dataset <b>FLAIR-INC</b> (x 3.5 patches). Only the FLAIR dataset is open at the moment.</li>
+    <li>input modalities : <b>RGB</b> (natural colours), <b>RGBI</b> (natural colours + infrared), <b>RGBIE</b> (natural colours + infrared + elevation)</li>
+    <li>model architecture : <b>resnet34_unet</b> (U-Net with a Resnet-34 encoder), <b>deeplab</b>, <b>fpn</b>, <b>mit</b></li>
+    <li>target class nomenclature : <b>12cl</b> (12 land cover classes) or <b>15cl</b> (15 land cover classes)</li>
+  </ul>
+</div>
+<br>
+
+<div style="border:1px solid black; padding:25px; background-color:#FDFFF4 ; padding-top:10px; padding-bottom:1px;">
+  <h1>FLAIR-INC_rgb_15cl_resnet34-deeplabv3</h1> 
+  <p>The general characteristics of this specific model <strong>FLAIR-INC_rgb_15cl_resnet34-deeplabv3</strong> are :</p>
+  <ul style="list-style-type:disc;">
+    <li>Trained with the FLAIR-INC dataset</li>
+    <li>Trained with the SegmentationModelsPytorch library</li>  
+    <li>RGB images (true colours)</li>
+    <li>DeepLabV3 with a Resnet-34 encoder</li>
+    <li>15 class nomenclature : [building, pervious surface, impervious surface, bare soil, water, coniferous, deciduous, brushwood, vineyard, herbaceous, agricultural land, plowed land, swimming pool, snow, greenhouse]</li>
+  </ul>
+</div>
+
+## Model Informations
+- **Code repository:** https://github.com/IGNF/FLAIR-1
+- **Paper:** https://arxiv.org/pdf/2211.12979.pdf
+- **Developed by:** IGN
+- **Compute infrastructure:** 
+    - software: python, pytorch-lightning
+    - hardware: HPC/AI resources provided by GENCI-IDRIS
+- **License:** Etalab 2.0
+
+---
+
+## Uses
+
+Although the model can be applied to other type of very high spatial earth observation images, it was initially developed to tackle the problem of classifying aerial images acquired on the French Territory.
+The product called ([BD ORTHO®](https://geoservices.ign.fr/bdortho)) has its own spatial and radiometric specifications. The model is not intended to be generic to other type of very high spatial resolution images but specific to BD ORTHO images. 
+Consequently, the model’s prediction would improve if the user images are similar to the original ones.
+
+_**Radiometry of input images**_ :
+The BD ORTHO input images are distributed in 8-bit encoding format per channel. When traning the model, input normalization was performed (see section **Training Details**). 
+It is recommended that the user apply the same type of input normalization while inferring the model.
+
+_**Multi-domain model**_ :
+The FLAIR-INC dataset that was used for training is composed of 75 radiometric domains. In the case of aerial images, domain shifts are frequent and are mainly due to : the date of acquisition of the aerial survey (from april to november), the spatial domain (equivalent to a french department administrative division) and downstream radiometric processing.
+By construction (sampling 75 domains) the model is robust to these shifts, and can be applied to any images of the ([BD ORTHO® product](https://geoservices.ign.fr/bdortho)).
+
+_**Land Cover classes of prediction**_ :
+The orginial class nomenclature of the FLAIR Dataset encompasses 19 classes (See the [FLAIR dataset](https://huggingface.co/datasets/IGNF/FLAIR) page for details).
+However 3 classes corresponding to uncertain labelisation (Mixed (16), Ligneous (17) and Other (19)) and 1 class with very poor labelling (Clear cut (15)) were desactivated during training.
+As a result, the logits produced by the model are of size 19x1, but classes n° 15, 16, 17 and 19 should appear at 0 in the logits and should not be present in the final argmax product.
+
+
+
+## Bias, Risks, Limitations and Recommendations
+
+_**Using the model on input images with other spatial resolution**_ :
+The FLAIR-INC_rgb_15cl_resnet34-deeplabv3 model was trained with fixed scale conditions. All patches used for training are derived from aerial images with 0.2 meters spatial resolution. Only flip and rotate augmentations were performed during the training process.  
+No data augmentation method concerning scale change was used during training. The user should pay attention that generalization issues can occur while applying this model to images that have different spatial resolutions.
+
+_**Using the model for other remote sensing sensors**_ :
+The FLAIR-INC_rgb_15cl_resnet34-deeplabv3 model was trained with aerial images of the ([BD ORTHO® product](https://geoservices.ign.fr/bdortho)) that encopass very specific radiometric image processing. 
+Using the model on other type of aerial images or satellite images may imply the use of transfer learning or domain adaptation techniques.
+
+_**Using the model on other spatial areas**_ :
+The FLAIR-INC_rgb_15cl_resnet34-deeplabv3 model was trained on patches reprensenting the French Metropolitan territory. 
+The user should be aware that applying the model to other type of landscapes may imply a drop in model metrics.  
+
+---
+
+## How to Get Started with the Model
+
+Visit ([https://github.com/IGNF/FLAIR-1](https://github.com/IGNF/FLAIR-1)) to use the model.
+Fine-tuning and prediction tasks are detailed in the README file.
+
+
+---
+
+## Training Details
+
+### Training Data
+
+218 400 patches of 512 x 512 pixels were used to train the **FLAIR-INC_rgb_15cl_resnet34-deeplabv3** model. 
+The train/validation split was performed patchwise to obtain a 80% / 20% distribution between train and validation. 
+Annotation was performed at the _zone_ level (~100 patches per _zone_). Spatial independancy between patches is guaranted as patches from the same _zone_ were assigned to the same set (TRAIN or VALIDATION). 
+The following number of patches were used for train and validation :
+| TRAIN set            | 174 700 patches    |
+| VALIDATION set       | 43 700 patchs      |
+
+
+
+
+
+### Training Procedure
+
+#### Preprocessing
+
+For traning the model, input normalization was performed to center-reduce (**a mean=0** and a **standard deviation = 1**, channel wise) the dataset.
+We used the statistics of TRAIN+VALIDATION for input normalization. It is recommended that the user apply the same type of input normalization. 
+
+Statistics of the TRAIN+VALIDATION set :
+
+| Modalities              | Mean (Train + Validation)       |Std    (Train + Validation)     |
+| ----------------------- | ----------- |----------- |
+| Red Channel (R)         | 105.08	    |52.17       |
+| Green Channel (G)       | 110.87      |45.38       |
+| Blue Channel (B)        | 101.82	    |44.00       |
+
+
+#### Training Hyperparameters
+
+```yaml
+- Model architecture: DeepLabV3 #(implementation from the [Segmentation Models Pytorch library](https://segmentation-modelspytorch.readthedocs.io/en/latest/docs/api.html#unet))
+- Encoder : Resnet-34 pre-trained with ImageNet
+- Augmentation :
+  - VerticalFlip(p=0.5)
+  - HorizontalFlip(p=0.5)
+  - RandomRotate90(p=0.5)
+- Input normalization (mean=0 | std=1):
+  - norm_means: [105.08, 110.87, 101.82]
+  - norm_stds: [52.17, 45.38, 44]
+- Seed: 2022
+- Batch size: 10
+- Number of epochs : 200
+- Early stopping : patience 30 and val_loss as monitor criterium 
+- Optimizer : SGD
+- Schaeduler : mode = "min", factor = 0.5, patience = 10, cooldown = 4, min_lr = 1e-7
+- Learning rate : 0.02
+- Class Weights : [1-building: 1.0 , 2-pervious surface: 1.0 , 3-impervious surface: 1.0 , 4-bare soil: 1.0 , 5-water: 1.0 , 6-coniferous: 1.0 , 7-deciduous: 1.0 , 8-brushwood: 1.0 , 9-vineyard: 1.0 , 10-herbaceous vegetation: 1.0 , 11-agricultural land: 1.0 , 12-plowed land: 1.0 , 13-swimming_pool: 1.0 , 14-snow: 1.0 , 15-clear cut: 0.0 , 16-mixed: 0.0 , 17-ligneous: 0.0 , 18-greenhouse: 1.0 , 19-other: 0.0]
+```
+
+#### Speeds, Sizes, Times
+
+The FLAIR-INC_rgb_15cl_resnet34-deeplabv3 model was trained on a HPC/AI resources provided by GENCI-IDRIS (Grant 2022-A0131013803). 
+16 V100 GPUs were used ( 4 nodes, 4 GPUS per node). With this configuration the approximate learning time is 6 minutes per epoch.
+
+FLAIR-INC_rgb_15cl_resnet34-deeplabv3  was obtained for num_epoch=22 with corresponding val_loss=0.60. 
+
+
+<div style="position: relative; text-align: center;">
+    <p style="margin: 0;">TRAIN loss</p>
+    <img src="FLAIR-INC_rgb_15cl_resnet34-deeplabv3_train-loss.png" alt="TRAIN loss" style="width: 60%; display: block; margin: 0 auto;"/>
+    <p style="margin: 0;">VALIDATION loss</p>
+    <img src="FLAIR-INC_rgb_15cl_resnet34-deeplabv3_val-loss.png" alt="VALIDATION loss" style="width: 60%; display: block; margin: 0 auto;"/>
+</div>
+
+
+
+## Evaluation
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+The evaluation was performed on a TEST set of 31 750 patches that are independant from the TRAIN and VALIDATION patches. They represent 15 spatio-temporal domains.
+The TEST set corresponds to the reunion of the TEST set of scientific challenges FLAIR#1 and FLAIR#2. See the [FLAIR challenge page](https://ignf.github.io/FLAIR/) for more details.
+
+The choice of a separate TEST set instead of cross validation was made to be coherent with the FLAIR challenges. 
+However the metrics for the Challenge were calculated on 12 classes and the TEST set acordingly. 
+As a result the _Snow_ class is absent from the TEST set.
+
+#### Metrics
+
+With the evaluation protocol, the **FLAIR-INC_rgb_15cl_resnet34-deeplabv3** have been evaluated to **OA= 74.887%** and **mIoU=58.093%**. 
+The _snow_ class is discarded from the average metrics.
+
+The following table give the class-wise metrics :
+
+|   Classes               |   IoU (%)    | Fscore (%)  | Precision (%)  | Recall (%)  |
+| ----------------------- | ----------|---------|---------|---------|
+|  building   		      |   74.671   |   85.499   |   82.639   |   88.565   |
+|  pervious_surface   	  |   52.890   |   69.187   |   73.606   |   65.269   |
+|  impervious_surface     |   69.452   |   81.972   |   81.036   |   82.930   |
+|  bare_soil   	          |   59.280   |   74.435   |   76.764   |   72.243   |
+|  water   		          |   82.144   |   90.197   |   88.130   |   92.363   |
+|  coniferous   	      |   55.443   |   71.336   |   75.030   |   67.988   |
+|  deciduous   	          |   68.459   |   81.277   |   77.756   |   85.132   |
+|  brushwood   	          |   28.215   |   44.012   |   61.283   |   34.336   |
+|  vineyard   		      |   77.997   |   87.639   |   84.523   |   90.994   |
+|  herbaceous   	      |   50.624   |   67.219   |   69.270   |   65.287   |
+|  agricultural_land      |   58.221   |   73.595   |   69.052   |   78.777   |
+|  plowed_land   	      |   43.881   |   60.996   |   60.336   |   61.671   |
+|  swimming_pool   	      |   38.207   |   55.289   |   59.539   |   51.606   |
+|  snow   		          |   _0.000_  |   _0.000_  |   _0.000_  |   _0.000_  |
+|  greenhouse   	      |   53.815   |   69.974   |   57.120   |   90.293   |
+| **average**             |  **58.093**    |  **72.331**  |  **72.577**  |  **73.390**  |
+
+
+
+
+
+
+
+
+The following illustration gives the resulting confusion matrix :
+* Top : normalised acording to columns, columns sum at 100% and the **precision** is on the diagonal of the matrix
+* Bottom : normalised acording to rows, rows sum at 100% and the **recall** is on the diagonal of the matrix 
+
+
+<div style="position: relative; text-align: center;">
+    <p style="margin: 0;">Normalized Confusion Matrix (precision)</p>
+    <img src="FLAIR-INC_rgb_15cl_resnet34-deeplabv3_confmat_norm-precision.png" alt="drawing" style="width: 70%; display: block; margin: 0 auto;"/>
+    <p style="margin: 0;">Normalized Confusion Matrix (recall)</p>
+    <img src="FLAIR-INC_rgb_15cl_resnet34-deeplabv3_confmat_norm-recall.png" alt="drawing" style="width: 70%; display: block; margin: 0 auto;"/>
+</div>
+
+
+
+### Results
+
+Samples of results
+
+
+---
+
+## Citation
+
+
+**BibTeX:**
+
+```
+@inproceedings{ign-flair,
+      title={FLAIR: a Country-Scale Land Cover Semantic Segmentation Dataset From Multi-Source Optical Imagery}, 
+      author={Anatol Garioud and Nicolas Gonthier and Loic Landrieu and Apolline De Wit and Marion Valette and Marc Poupée and Sébastien Giordano and Boris Wattrelos},
+      year={2023},
+      booktitle={Advances in Neural Information Processing Systems (NeurIPS) 2023},
+      doi={https://doi.org/10.48550/arXiv.2310.13336},
+}
+```
+
+
+**APA:**
+```
+Anatol Garioud, Nicolas Gonthier, Loic Landrieu, Apolline De Wit, Marion Valette, Marc Poupée, Sébastien Giordano and Boris Wattrelos. 2023. 
+FLAIR: a Country-Scale Land Cover Semantic Segmentation Dataset From Multi-Source Optical Imagery. (2023).
+In proceedings of Advances in Neural Information Processing Systems (NeurIPS) 2023.
+DOI: https://doi.org/10.48550/arXiv.2310.13336
+```
+
+## Contact : flair@ign.fr

models/land_cover/flair-rgb-15cl-deeplabv3/confmat.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d5d0c9807283cc4bec2cfdf8794aa6175df82735d0a8e31d177ff9ffc5c7cff
+size 3016

models/landcover/deeplabv3_landcover_4c.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c82806dfe6a582ba26642427ef5533595981ec2bb6d4d3a3c4a3f06ac4753c4
+size 12336439

models/landcover/provenance.json

@@ -0,0 +1,10 @@
+{
+  "repository": "PUTvision/qgis-plugin-deepness",
+  "revision": "905ac02affa45b229c45bf8bc0a082a3a7700d29",
+  "url": "https://raw.githubusercontent.com/PUTvision/qgis-plugin-deepness/905ac02affa45b229c45bf8bc0a082a3a7700d29/examples/deeplabv3_segmentation_landcover/deeplabv3_landcover_4c.onnx",
+  "sha256": "3c82806dfe6a582ba26642427ef5533595981ec2bb6d4d3a3c4a3f06ac4753c4",
+  "bytes": 12336439,
+  "source_documentation": "https://qgis-plugin-deepness.readthedocs.io/en/latest/example/example_segmentation_landcover.html",
+  "license": "GPL-3.0 repository; model-specific license not separately documented",
+  "status": "evaluation_only"
+}

models/manifests/models.yaml

@@ -0,0 +1,419 @@
+version: 2.0.0
+storage_root: models
+models:
+  water_s2_surface:
+    model_id: giswqs/s2-water-unetplusplus-efficientnet-b4
+    display_name: Sentinel-2 surface water UNet++
+    path: water/s2-water-unetplusplus-efficientnet-b4
+    category: water
+    priority: six_band_s2
+    policy: PRODUCTION
+    license: CC-BY-4.0
+    source_url: https://huggingface.co/giswqs/s2-water-unetplusplus-efficientnet-b4
+    revision: efb6f636480188f48779b41f2a94bb941fd7b900
+    task:
+    - water_segmentation
+    - water_detection
+    - water_area
+    modality:
+    - multispectral_s2
+    expected_channels:
+    - B02
+    - B03
+    - B04
+    - B08
+    - B11
+    - B12
+    input_size:
+    - 512
+    - 512
+    - 6
+    output_type: pixel_water_probability
+    checkpoint: model.pth
+    notes:
+      compatibility: Named Sentinel-2 bands on a georeferenced 8-30 metre grid. Never
+        substitute RGB channels.
+      sha256: fb8ae300b8b8e929fa764d72ac476a419e8eb9f121e52cce6c0c8816a48d487e
+  land_deepness:
+    model_id: PUTvision/Deepness-LandCoverAI-DeepLabV3plus
+    display_name: Deepness aerial land cover
+    path: landcover
+    category: landcover
+    priority: primary_aerial
+    policy: PRODUCTION
+    license: GPL-3.0 repository; model-specific terms not separately documented
+    source_url: https://qgis-plugin-deepness.readthedocs.io/en/latest/example/example_segmentation_landcover.html
+    revision: 905ac02affa45b229c45bf8bc0a082a3a7700d29
+    task:
+    - dense_landcover_segmentation
+    modality:
+    - rgb_vhr
+    expected_channels:
+    - red
+    - green
+    - blue
+    input_size:
+    - 512
+    - 512
+    - 3
+    output_type: five_class_pixel_probabilities
+    checkpoint: deeplabv3_landcover_4c.onnx
+    notes:
+      compatibility: Verified 0.15-0.60 metre RGB pixels; geography unvalidated.
+      class_map:
+      - unknown
+      - building
+      - woodland
+      - water
+      - road
+      sha256: 3c82806dfe6a582ba26642427ef5533595981ec2bb6d4d3a3c4a3f06ac4753c4
+  building_satellite:
+    model_id: merve/rf-detr-seg-satellite-buildings
+    display_name: RF-DETR satellite building instances
+    path: buildings/rf-detr-seg-satellite-buildings
+    category: buildings
+    priority: default_satellite
+    policy: PRODUCTION
+    license: Apache-2.0
+    source_url: https://huggingface.co/merve/rf-detr-seg-satellite-buildings
+    revision: 05b80dce9a57701724ad6a0fc052827c8b724257
+    checkpoint: model.safetensors
+    task:
+    - building_detection
+    - building_footprint
+    - building_count
+    - building_grounding
+    modality:
+    - rgb
+    expected_channels:
+    - red
+    - green
+    - blue
+    input_size:
+    - 432
+    - 432
+    - 3
+    output_type: instance_masks_and_scores
+    notes:
+      compatibility: VHR RGB satellite imagery; target geography unvalidated; rejects
+        pixels coarser than 1.5 m.
+      selection: Best of three measured satellite candidates on 16 previously inspected
+        images. Not independently validated for arbitrary aerial images.
+      sha256: f254c1400f780f7ea72a6bc588a2150bf8b4d8845ded7269655e26a275f41ac7
+  building_primary:
+    model_id: satquery_buildings_bundle
+    display_name: SatQuery Satlas Buildings v1
+    base_model: AI2 SatlasPretrain Aerial_SwinB_SI
+    version: 1.0.0
+    training_dataset: SpaceNet 2 (Vegas/Shanghai train, Paris validation, Khartoum
+      test)
+    policy: CANDIDATE
+    threshold_file: thresholds.json
+    path: finetuned/buildings/satquery_buildings_v1
+    category: buildings
+    priority: explicit_bundle_override
+    status: READY
+    license: CC-BY-4.0
+    source_url: models/satquery_buildings_bundle
+    task:
+    - building_detection
+    - building_footprint
+    - building_count
+    - building_grounding
+    modality:
+    - rgb
+    expected_channels:
+    - red
+    - green
+    - blue
+    input_size:
+    - 512
+    - 512
+    - 3
+    output_type: footprint_and_boundary_probabilities
+    notes:
+      default_window_size: '512'
+      default_stride: '384'
+      default_overlap: '128'
+      default_foot_thr: '0.6'
+      default_boundary_thr: '0.6'
+      method: Hann blend then global watershed
+      backbone: Aerial_SwinB_SI
+      fine_tuned_on: SpaceNet 2
+    checkpoint: satquery_buildings_state_dict.pt
+    health_status: READY
+  building_secondary:
+    model_id: geobase/building-detection
+    path: buildings/geobase-building-detection
+    category: buildings
+    priority: secondary
+    status: READY
+    license: CC-BY-4.0
+    source_url: https://huggingface.co/geobase/building-detection
+    task:
+    - building_validation
+    - secondary_building_check
+    modality:
+    - rgb
+    expected_channels: &id001
+    - red
+    - green
+    - blue
+    input_size:
+    - 256
+    - 256
+    - 3
+    output_type: building_detection_logits
+    notes:
+      role: Secondary/validation building model. Do not make default until benchmark
+        proves improvement.
+    display_name: Building Secondary
+    download_status: UNVERIFIED
+    channel_order: *id001
+    checkpoint: onnx/model_quantized.onnx
+    revision: null
+    health_status: READY
+    normalization:
+      type: checkpoint_native
+    input_range: checkpoint_native
+    expected_resolution: see model card
+    class_map: null
+    threshold_file: null
+    device_preference: auto
+    precision: float32
+    fallbacks: []
+  water_finetuned:
+    model_id: satquery_water_bundle
+    display_name: SatQuery Satlas Sentinel-2 Water v1
+    base_model: AI2 SatlasPretrain Sentinel2_SwinB_SI_MS
+    version: 1.0.0
+    training_dataset: Sen1Floods11 with event-exclusive splits and shadow hard negatives
+    policy: PRODUCTION
+    threshold_file: thresholds.json
+    path: satquery_water_bundle
+    category: water
+    priority: primary_water_s2
+    status: READY
+    task:
+    - water_segmentation
+    - water_detection
+    - water_area
+    - water_grounding
+    modality:
+    - multispectral_s2
+    expected_channels:
+    - B02
+    - B03
+    - B04
+    - B05
+    - B06
+    - B07
+    - B08
+    - B10
+    - B11
+    - B12
+    input_size:
+    - 512
+    - 512
+    - 10
+    output_type: pixel_surface_water_probability
+    notes:
+      compatibility: Exact Sentinel-2 bands only; 9-band Satlas backbone plus a six-feature
+        spectral branch. Never use on RGB or SAR.
+      backbone_band_order:
+      - B04
+      - B03
+      - B02
+      - B05
+      - B06
+      - B07
+      - B08
+      - B11
+      - B12
+      spectral_features:
+      - NDWI
+      - MNDWI
+      - NDVI
+      - AWEI_shadow
+      - visible_brightness
+      - B10_cirrus
+      shadow_contract: The checkpoint predicts surface water only. Shadow is produced
+        independently by the illumination engine.
+    health_status: READY
+  land_rgb:
+    model_id: IGNF/FLAIR-INC_rgb_15cl_resnet34-deeplabv3
+    path: land_cover/flair-rgb-15cl-deeplabv3
+    category: land_cover
+    priority: primary
+    status: READY
+    license: Etalab-2.0
+    source_url: https://huggingface.co/IGNF/FLAIR-INC_rgb_15cl_resnet34-deeplabv3
+    task:
+    - land_cover_segmentation
+    - pixel_land_cover_rgb
+    modality:
+    - rgb
+    expected_channels: &id002
+    - red
+    - green
+    - blue
+    input_size:
+    - 512
+    - 512
+    - 3
+    output_type: 15_class_pixel_probabilities
+    notes:
+      compatibility: High-resolution RGB imagery only (0.2m-0.5m GSD). Do NOT apply
+        to SAR or coarse multispectral tensors.
+      classes:
+      - building
+      - pervious_surface
+      - impervious_surface
+      - bare_soil
+      - water
+      - coniferous
+      - deciduous
+      - brushwood
+      - vineyard
+      - herbaceous_vegetation
+      - agricultural_land
+      - plowed_land
+      - swimming_pool
+      - greenhouse
+      - other
+    display_name: Land Rgb
+    download_status: UNVERIFIED
+    channel_order: *id002
+    checkpoint: FLAIR-INC_rgb_15cl_resnet34-deeplabv3_weights.pth
+    revision: null
+    health_status: READY
+    normalization:
+      type: checkpoint_native
+    input_range: checkpoint_native
+    expected_resolution: see model card
+    class_map: null
+    threshold_file: null
+    device_preference: auto
+    precision: float32
+    fallbacks: []
+  satlas_aerial_swinb_si:
+    model_id: allenai/satlas-pretrain
+    display_name: Satlas Aerial Swinb Si
+    path: pretrained/buildings/satlas-aerial-swinb-si
+    category: buildings
+    priority: foundation
+    download_status: UNVERIFIED
+    license: Apache-2.0
+    source_url: https://github.com/allenai/satlaspretrain_models
+    task:
+    - building_backbone
+    modality:
+    - rgb_vhr
+    expected_channels: &id003
+    - red
+    - green
+    - blue
+    channel_order: *id003
+    input_size:
+    - 512
+    - 512
+    - 3
+    output_type: multiscale_features_no_task_head
+    checkpoint: aerial_swinb_si.pth
+    revision: b1f8de04dacf9f9ed9640fe32484947710b47ede
+    notes:
+      checkpoint_id: Aerial_SwinB_SI
+      requires_task_head: true
+    health_status: READY
+    normalization:
+      type: checkpoint_native
+    input_range: checkpoint_native
+    expected_resolution: see model card
+    class_map: null
+    threshold_file: null
+    device_preference: auto
+    precision: float32
+    fallbacks: []
+    status: READY
+  land_flair_hub:
+    model_id: IGNF/FLAIR-HUB_LC-A_RGB_swinbase-upernet
+    display_name: Land Flair Hub
+    path: pretrained/landcover/flair-hub-rgb-swinbase-upernet
+    category: landcover
+    priority: baseline
+    download_status: UNVERIFIED
+    license: Etalab-2.0
+    source_url: https://huggingface.co/IGNF/FLAIR-HUB_LC-A_RGB_swinbase-upernet
+    task:
+    - dense_landcover_segmentation
+    modality:
+    - rgb_vhr
+    expected_channels: &id004
+    - red
+    - green
+    - blue
+    channel_order: *id004
+    input_size:
+    - 512
+    - 512
+    - 3
+    output_type: 19_channel_pixel_logits_15_evaluated_classes
+    checkpoint: FLAIR-HUB_LC-A_RGB_swinbase-upernet.safetensors
+    revision: 88190fc26a325509a78e4a7289e2ab6a8f5d58a4
+    notes:
+      class_map_source: configs_train/config_supervision.yaml
+      ignored_class_ids:
+      - 15
+      - 16
+      - 17
+      - 18
+    health_status: READY
+    normalization:
+      type: checkpoint_native
+    input_range: checkpoint_native
+    expected_resolution: see model card
+    class_map: null
+    threshold_file: null
+    device_preference: auto
+    precision: float32
+    fallbacks: []
+    status: READY
+  landcover_candidate:
+    model_id: satquery_landcover_v1
+    display_name: SatQuery Land Cover v1 candidate
+    path: finetuned/landcover/satquery_landcover_v1
+    task:
+    - dense_landcover_segmentation
+    modality:
+    - rgb_vhr
+    expected_channels:
+    - red
+    - green
+    - blue
+    policy: CANDIDATE
+    priority: candidate
+    health_status: READY
+    notes:
+      reason: Bundle README reports a bounded toy validation run; independent target-domain
+        promotion gate has not passed.
+    status: READY
+  water_shadow_rgb:
+    model_id: satquery_water_shadow_v1
+    display_name: SatQuery RGB Water and Shadow v1
+    path: finetuned/water/satquery_water_shadow_v1
+    task:
+    - water_segmentation
+    - shadow_segmentation
+    modality:
+    - rgb_vhr
+    expected_channels:
+    - red
+    - green
+    - blue
+    policy: CANDIDATE
+    priority: primary_when_validated
+    health_status: READY
+    notes:
+      reason: The existing SatQuery water bundle is Sentinel-2 water only; it is not
+        this checkpoint.
+    status: READY

models/pretrained/buildings/satlas-aerial-swinb-si/aerial_swinb_si.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8abf21cadaab2a71859e9d172bec4e5d6b7cc9710882a2f6a4d8aae826db036
+size 504439674

models/pretrained/buildings/satlas-aerial-swinb-si/satlas_metadata.json

@@ -0,0 +1,9 @@
+{
+  "checkpoint_id": "Aerial_SwinB_SI",
+  "official_url": "https://huggingface.co/allenai/satlas-pretrain/resolve/main/aerial_swinb_si.pth?download=true",
+  "repository_revision": "b1f8de04dacf9f9ed9640fe32484947710b47ede",
+  "package": "satlaspretrain-models",
+  "input_channels": 3,
+  "multi_image": false,
+  "task_head": "none; foundation features only"
+}

models/pretrained/buildings/satlas-aerial-swinb-si/satquery_download_metadata.json

@@ -0,0 +1,5 @@
+{
+  "repository": "allenai/satlas-pretrain",
+  "revision": "b1f8de04dacf9f9ed9640fe32484947710b47ede",
+  "download_date": "2026-09-23T15:18:43.138074+00:00"
+}

models/pretrained/buildings/satlas-aerial-swinb-si/satquery_model_health.json

@@ -0,0 +1,35 @@
+{
+  "inference_verified": true,
+  "checkpoint_sha256": "f8abf21cadaab2a71859e9d172bec4e5d6b7cc9710882a2f6a4d8aae826db036",
+  "verified_at": "2026-09-23T19:03:31.047526+00:00",
+  "checkpoint_id": "Aerial_SwinB_SI",
+  "output_shapes": [
+    [
+      1,
+      128,
+      128,
+      128
+    ],
+    [
+      1,
+      256,
+      64,
+      64
+    ],
+    [
+      1,
+      512,
+      32,
+      32
+    ],
+    [
+      1,
+      1024,
+      16,
+      16
+    ]
+  ],
+  "device": "cuda",
+  "peak_allocated_vram_bytes": 469159936,
+  "role": "foundation backbone; no building task head"
+}

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/.gitattributes

@@ -0,0 +1,39 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+model_utils/AerialROI.png filter=lfs diff=lfs merge=lfs -text
+model_utils/FLAIR-HUB_split1_classesfreq.png filter=lfs diff=lfs merge=lfs -text
+model_utils/FLAIR-HUB_LC-A_RVB_swinbase-upernet_inferenceROI.png filter=lfs diff=lfs merge=lfs -text
+model_utils/FLAIR-HUB_LC-A_RGB_swinbase-upernet_inferenceROI.png filter=lfs diff=lfs merge=lfs -text

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/README.md

@@ -0,0 +1,433 @@
+---
+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
+---
+
+<div style="font-family:sans-serif; color:black; background-color:#F8F5F5; padding:25px; border-radius:10px; margin:auto; border:0px; ">
+
+  <!-- Collection Section -->
+  <div style="background:#FFFFFF; color:black; padding:20px; border-radius:8px; box-shadow:0 2px 5px rgba(0,0,0,0.05); margin-bottom:20px;">
+    <h1 style="margin-top:0; color:black;">🌐 FLAIR-HUB Model Collection</h1>
+    <ul style="padding-left:0; list-style:none; line-height:1.6; margin:0;">
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+      <b>Trained on</b>: <span style="color:black;">FLAIR-HUB dataset</span> 
+      <a href="https://huggingface.co/datasets/IGNF/FLAIR-HUB" target="_blank" style="margin-left:5px;">🔗</a>
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Available modalities</b>: Aerial images, SPOT images, Topographic info, Sentinel-2 yearly time-series, Sentinel-1 yearly time-series, Historical aerial images
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Encoders</b>: ConvNeXTV2, Swin (Tiny, Small, Base, Large)
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Decoders</b>: UNet, UPerNet
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Tasks</b>: Land-cover mapping (LC), Crop-type mapping (LPIS)
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Class nomenclature</b>: 15 classes for LC, 23 classes for LPIS
+      </li>
+    </ul>
+      <table border="1" style="border-collapse: collapse; width:100%; margin-bottom:15px; table-layout: fixed;">
+      <thead>
+        <tr>
+          <th style="padding:1px;"></th>
+          <th colspan="2" style="padding:1px; text-align:center; font-weight:bold; color:black;">
+            Model task
+          </th>
+          <th colspan="6" style="padding:1px; text-align:center; font-weight:bold; color:black;">
+            Model inputs
+          </th>
+        </tr>
+        <tr>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🆔<br>Model ID</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🗺️<br>Land-cover</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🌾<br>Crop-types</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🛩️<br>Aerial</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">⛰️<br>Elevation</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🛰️<br>SPOT</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🛰️<br>S2 t.s.</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🛰️<br>S1 t.s.</th>
+          <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">🛩️<br>Historical</th>
+        </tr>
+      </thead>
+      <tbody>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-A</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-B</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr> 
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-D</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-F</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-G</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-I</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LC-L</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LPIS-A</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LPIS-F</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LPIS-I</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+        <tr>
+          <td style="padding:1px; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">LPIS-J</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">✓</td>
+          <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;"></td>
+        </tr>
+      </tbody>
+    </table>
+  </div>
+
+  <!-- Model-Specific Section -->
+  <div style="border:1px solid black; color:black; padding:25px; background-color:#FDFFF4; border-radius:8px; box-shadow:0 2px 5px rgba(0,0,0,0.05);">
+    <h2 style="margin-top:0; color:black;">🔍 Model: FLAIR-HUB_LC-A_RVB_swinbase-upernet</h2>
+    <ul style="padding-left:0; list-style:none; line-height:1.6; margin:0;">
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Encoder</b>: <i>swin_base_patch4_window12_384</i>
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Decoder</b>: <i>upernet</i>
+      </li>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Metrics</b>:
+      </li>  
+<table border="1" style="border-collapse: collapse; width:100%; margin-bottom:15px; table-layout: fixed;">
+<thead>
+  <tr>
+    <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">mIoU</th>
+    <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">O.A.</th>
+    <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">F-score</th>
+    <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">Precision</th>
+    <th style="padding:1px; text-align:center; color:black; width:5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">Recall</th>
+  </tr>
+</thead>
+<tr>
+  <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">64.05%</td>
+  <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">76.78%</td>
+  <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">76.88%</td>
+  <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">77.71%</td>
+  <td style="padding:1px; text-align:center; width5%; white-space:nowrap; overflow:hidden; text-overflow:ellipsis;">76.59%</td>
+</tr>
+</table>
+      <li>
+        <span style="display:inline-block; width:10px; height:10px; background:#555; border-radius:2px; margin-right:10px; box-shadow:1px 1px 2px rgba(0,0,0,0.2); vertical-align:middle;"></span>
+        <b>Params.</b>: <i>89.4</i>
+      </li>
+    </ul>
+  </div>
+</div>
+
+
+---
+
+## General Informations
+- **Contact:** flair@ign.fr
+- **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
+
+```yaml
+- 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
+
+```yaml
+- Train patches: 152225
+- Validation patches: 38175
+- Test patches: 50700
+```
+
+<div style="position: relative; text-align: center;">
+    <img src="./model_utils/FLAIR-HUB_split1_classesfreq.png" alt="Classes distribution." style="width: 100%; display: block; margin: 0 auto;"/>
+</div>
+
+---
+
+### Training Logging
+
+<div style="position: relative; text-align: center;">
+    <img src="./model_utils/FLAIR-HUB_LC-A_RGB_swinbase-upernet_logs.png" alt="Training logging." style="width: 100%; display: block; margin: 0 auto;"/>
+</div>
+
+---
+
+
+## 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
+
+<div style="display: flex; justify-content: center; text-align: center; gap: 20px;">
+
+  <div style="flex: 1;">
+    <p style="margin: 0;">Aerial ROI</p>
+    <img src="./model_utils/AerialROI.png" alt="AERIAL" style="width: 100%; display: block;" />
+  </div>
+  <div style="flex: 1;">
+    <p style="margin: 0;">Inference ROI</p>
+    <img src="./model_utils/FLAIR-HUB_LC-A_RGB_swinbase-upernet_inferenceROI.png" alt="INFERENCE" style="width: 100%; display: block;" />
+  </div>
+</div>
+
+
+---
+
+
+## Cite
+
+
+**BibTeX:**
+
+```
+@article{GARIOUD2026271,
+title = {FLAIR-HUB: Large-scale multimodal dataset for land cover and crop mapping},
+author = {Anatol Garioud and Sébastien Giordano and Nicolas David and Nicolas Gonthier},
+journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
+volume = {237},
+pages = {271-300},
+year = {2026},
+issn = {0924-2716},
+doi = {https://doi.org/10.1016/j.isprsjprs.2026.04.017},
+url = {https://www.sciencedirect.com/science/article/pii/S0924271626001899},
+}
+```
+
+**APA:**
+```
+Anatol Garioud, Sébastien Giordano, Nicolas David, Nicolas Gonthier. 
+FLAIR-HUB: Large-scale Multimodal Dataset for Land Cover and Crop Mapping.
+ISPRS Journal of Photogrammetry and Remote Sensing, Volume 237, 2026.
+DOI: https://doi.org/10.1016/j.isprsjprs.2026.04.017
+```

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/configs_train/config_modalities.yaml

@@ -0,0 +1,65 @@
+modalities: 
+
+    inputs:
+        AERIAL_RGBI       : True
+        AERIAL-RLT_PAN    : False
+        DEM_ELEV          : False
+        SPOT_RGBI         : False
+        SENTINEL2_TS      : False 
+        SENTINEL1-ASC_TS  : False
+        SENTINEL1-DESC_TS : False
+        
+    inputs_channels:
+        AERIAL_RGBI       : [1,2,3] 
+        SPOT_RGBI         : 
+        SENTINEL2_TS      : 
+        SENTINEL1-ASC_TS  : 
+        SENTINEL1-DESC_TS : 
+
+    aux_loss: 
+        AERIAL_RGBI       : False
+        AERIAL-RLT_PAN    : False
+        DEM_ELEV          : False
+        SPOT_RGBI         : False
+        SENTINEL2_TS      : False
+        SENTINEL1-ASC_TS  : False
+        SENTINEL1-DESC_TS : False
+
+    aux_loss_weight: 1 # multiplier before adding to main loss
+
+    modality_dropout: # between 0 (no dropout) and 1 (complete systematic dropout)
+        AERIAL_RGBI       : 0
+        AERIAL-RLT_PAN    : 0
+        DEM_ELEV          : 0
+        SPOT_RGBI         : 0
+        SENTINEL2_TS      : 0
+        SENTINEL1-ASC_TS  : 0
+        SENTINEL1-DESC_TS : 0
+
+    pre_processings: 
+        filter_sentinel2: True 
+        filter_sentinel2_max_cloud : 1 # [0-100]
+        filter_sentinel2_max_snow : 1 # [0-100]
+        filter_sentinel2_max_frac_cover : 0.05 # [0-1]
+        temporal_average_sentinel2 : False  # possible : False, monthly, semi-monthly
+        temporal_average_sentinel1 : False
+
+        calc_elevation : True
+        calc_elevation_stack_dsm : True 
+        
+        use_augmentation: False
+
+    normalization: 
+        norm_type : custom   # possible :  custom, scaling, without
+
+        AERIAL_RGBI_means : [105.66, 111.35, 102.18] 
+        AERIAL_RGBI_stds  : [52.23, 45.62, 44.30]
+
+        AERIAL-RLT_PAN_means : [125.92]
+        AERIAL-RLT_PAN_stds  : [38.45]
+
+        SPOT_RGBI_means : [433.26, 508.75, 467.77, 1137.03]
+        SPOT_RGBI_stds  : [312.76, 284.61, 226.02, 543.11]
+
+        DEM_ELEV_means : [311.06, 311.06] # use same for both DSM/DTM to allow keeping differences of elevation
+        DEM_ELEV_stds  : [537.55, 537.55] # use same for both DSM/DTM to allow keeping differences of elevation

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/configs_train/config_models.yaml

@@ -0,0 +1,23 @@
+models: 
+
+    monotemp_model: # encoder-decoder from SMP
+
+        arch: swin_base_patch4_window12_384-upernet
+        new_channels_init_mode:
+
+    multitemp_model: 
+    
+        ref_date: '05-15' # defined for whole dataset    
+        encoder_widths: [64, 64, 64, 128] # last must be equivalent to decoder_widths
+        decoder_widths: [32, 32, 64, 128] # last must be equivalent to encoder_widths
+        out_conv: [32, 19]
+        str_conv_k: 3
+        str_conv_s: 1
+        str_conv_p: 1
+        agg_mode: "att_group"
+        encoder_norm: "group"
+        n_head: 16
+        d_model: 256
+        d_k: 4
+        pad_value: 0
+        padding_mode: "reflect"

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/configs_train/config_supervision.yaml

@@ -0,0 +1,39 @@
+labels: 
+  - AERIAL_LABEL-COSIA
+
+labels_configs:
+    AERIAL_LABEL-COSIA:
+        task_weight: 1
+        value_name:
+            0  : 'building'
+            1  : 'greenhouse'
+            2  : 'swimming_pool'
+            3  : 'impervious surface'
+            4  : 'pervious surface'
+            5  : 'bare soil'
+            6  : 'water'
+            7  : 'snow'
+            8  : 'herbaceous vegetation'
+            9  : 'agricultural land'
+            10 : 'plowed land'
+            11 : 'vineyard'
+            12 : 'deciduous'
+            13 : 'coniferous'
+            14 : 'brushwood'
+            15 : 'clear cut'
+            16 : 'ligneous'
+            17 : 'mixed'
+            18 : 'undefined'
+        value_weights: 
+            default: 1
+            default_exceptions:
+                15: 0
+                16: 0
+                17: 0
+                18: 0
+            per_modality_exceptions:
+                AERIAL_RGBI: 
+                SENTINEL2_TS: 
+                SENTINEL1-ASC_TS: 
+                SENTINEL1-DESC_TS:
+

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/configs_train/config_task.yaml

@@ -0,0 +1,61 @@
+# SLURM
+
+paths :   
+
+    out_folder: '../'
+    out_model_name: 'FLAIR-HUB_LC-A_RGB_swinbase-upernet'
+
+    train_csv: '../TRAIN_FLAIR-INC.csv'
+    val_csv: '../VALID_FLAIR-INC.csv'
+    test_csv: '../TEST_FLAIR-INC.csv'
+    global_mtd_folder: '../GLOBAL_ALL_MTD/'
+    ckpt_model_path: ''
+
+
+tasks: 
+
+    train: True
+    train_tasks: 
+        init_weights_only_from_ckpt: False
+        resume_training_from_ckpt: False
+
+    predict: True
+    write_files: True
+    georeferencing_output: True
+
+    metrics_only: False
+
+hyperparams:
+
+    num_epochs: 150
+    batch_size: 5
+    seed: 2025
+    learning_rate: 0.00005
+
+    optimizer: adamw  #sgd, adam, adamw
+    optim_weight_decay: 0.01
+    optim_betas: [0.9, 0.999]
+
+    scheduler: one_cycle_lr #  [one_cycle_lr, reduce_on_plateau, cycle_then_plateau]
+    warmup_fraction: 0.2 #if using one_cycle_lr. [0-1]
+    plateau_patience: 5
+
+hardware:
+    accelerator: 'gpu'
+    num_nodes: 6
+    gpus_per_node: 4
+    strategy: 'ddp_find_unused_parameters_true'
+    num_workers: 10
+
+saving:   
+    ckpt_save_also_last: True
+    ckpt_weights_only: False
+    ckpt_monitor: 'val_miou'
+    ckpt_monitor_mode: 'max'
+    ckpt_earlystopping_patience: 20
+    cp_csv_and_conf_to_output: True
+    enable_progress_bar: True
+    progress_rate: 10
+
+    ckpt_verbose: True
+    verbose_config: False

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/satquery_download_metadata.json

@@ -0,0 +1,5 @@
+{
+  "repository": "IGNF/FLAIR-HUB_LC-A_RGB_swinbase-upernet",
+  "revision": "88190fc26a325509a78e4a7289e2ab6a8f5d58a4",
+  "download_date": "2026-09-23T15:19:56.706918+00:00"
+}

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/satquery_model_health.json

@@ -0,0 +1,36 @@
+{
+  "inference_verified": true,
+  "checkpoint_sha256": "a995a226065b176c29f0a66a4937cc8bcea5741887b3f8c17812c430f75f4e3c",
+  "verified_at": "2026-09-23T19:03:34.945629+00:00",
+  "output_shape": [
+    1,
+    19,
+    512,
+    512
+  ],
+  "class_map": [
+    "building",
+    "greenhouse",
+    "swimming_pool",
+    "impervious surface",
+    "pervious surface",
+    "bare soil",
+    "water",
+    "snow",
+    "herbaceous vegetation",
+    "agricultural land",
+    "plowed land",
+    "vineyard",
+    "deciduous",
+    "coniferous",
+    "brushwood",
+    "clear cut",
+    "ligneous",
+    "mixed",
+    "undefined"
+  ],
+  "device": "cuda",
+  "peak_allocated_vram_bytes": 684190720,
+  "preprocessing": "official AERIAL_RGBI custom mean/std",
+  "runtime": "segmentation-models-pytorch==0.4.0"
+}

models/pretrained/landcover/flair-hub-rgb-swinbase-upernet/upstream/configs/train/config_models.yaml

@@ -0,0 +1,23 @@
+models: 
+
+    monotemp_model: 
+
+        arch: 'swin_base_patch4_window12_384-upernet' 
+        new_channels_init_mode: 'random' # [copy_first / copy_second / copy_third / random]
+
+    multitemp_model: 
+    
+        ref_date: '05-15' # defined for whole dataset    
+        encoder_widths: [64, 64, 64, 128] # last must be equivalent to decoder_widths
+        decoder_widths: [32, 32, 64, 128] # last must be equivalent to encoder_widths
+        out_conv: [32, 19]
+        str_conv_k: 3
+        str_conv_s: 1
+        str_conv_p: 1
+        agg_mode: "att_group"
+        encoder_norm: "group"
+        n_head: 16
+        d_model: 256
+        d_k: 4
+        pad_value: 0
+        padding_mode: "reflect"