metadata

license: etalab-2.0
tags:
  - segmentation
  - pytorch
  - aerial imagery
  - landcover
  - IGN
model-index:
  - name: FLAIR-INC_rgb_15cl_unetresnet34
    results:
      - task:
          type: semantic-segmentation
        dataset:
          name: IGNF/FLAIR/
          type: earth-observation-dataset
        metrics:
          - name: mIoU
            type: mIoU
            value: 59.013
          - name: Overall Accuracy
            type: OA
            value: 75.482
          - name: Fscore
            type: Fscore
            value: 72.926
          - name: Precision
            type: Precision
            value: 75.127
          - name: Recall
            type: Recall
            value: 71.993
          - name: IoU Buildings
            type: IoU
            value: 78.579
          - name: IoU Pervious surface
            type: IoU
            value: 52.999
          - name: IoU Impervious surface
            type: IoU
            value: 71.921
          - name: IoU Bare soil
            type: IoU
            value: 60.599
          - name: IoU Water
            type: IoU
            value: 84.561
          - name: IoU Coniferous
            type: IoU
            value: 56.066
          - name: IoU Deciduous
            type: IoU
            value: 69.527
          - name: IoU Brushwood
            type: IoU
            value: 27.868
          - name: IoU Vineyard
            type: IoU
            value: 75.716
          - name: IoU Herbaceous vegetation
            type: IoU
            value: 51.938
          - name: IoU Agricultural land
            type: IoU
            value: 57.725
          - name: IoU Plowed land
            type: IoU
            value: 40.911
          - name: IoU Swimming pool
            type: IoU
            value: 37.081
          - name: IoU Greenhouse
            type: IoU
            value: 60.688
pipeline_tag: image-segmentation

FLAIR model collection

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 BD ORTHO® product). The distributed pre-trained models differ in their :

dataset for training : FLAIR dataset or the increased version of this dataset FLAIR-INC (x 3.5 patches). Only the FLAIR dataset is open at the moment.
input modalities : RGB (natural colours), RGBI (natural colours + infrared), RGBIE (natural colours + infrared + elevation)
model architecture : resnet34_unet (U-Net with a Resnet-34 encoder), deeplab, fpn, mit
target class nomenclature : 12cl (12 land cover classes) or 15cl (15 land cover classes)

FLAIR-INC_rgb_15cl_resnet34-unet

The general characteristics of this specific model FLAIR-INC_rgb_15cl_resnet34-unet are :

Trained with the FLAIR-INC dataset
Trained with the SegmentationModelsPytorch library
RGB images (true colours)
U-Net with a Resnet-34 encoder
15 class nomenclature : [building, pervious surface, impervious surface, bare soil, water, coniferous, deciduous, brushwood, vineyard, herbaceous, agricultural land, plowed land, swimming pool, snow, greenhouse]

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®) 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).

Land Cover classes of prediction : The orginial class nomenclature of the FLAIR Dataset encompasses 19 classes (See the FLAIR dataset 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-unet 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-unet model was trained with aerial images of the (BD ORTHO® product) 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-unet 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) 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-unet 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

- Model architecture: Unet #(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-unet 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-unet was obtained for num_epoch=72 with corresponding val_loss=0.58.

TRAIN loss

VALIDATION loss

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 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-unet have been evaluated to OA= 75.482% and mIoU=59.013%. The snow class is discarded from the average metrics.

The following table give the class-wise metrics :

Classes	IoU (%)	Fscore (%)	Precision (%)	Recall (%)
building	78.579	88.004	88.078	87.931
pervious_surface	52.999	69.280	71.442	67.245
impervious_surface	71.921	83.668	83.877	83.460
bare_soil	60.599	75.466	73.608	77.420
water	84.561	91.635	91.687	91.583
coniferous	56.066	71.849	76.857	67.454
deciduous	69.527	82.025	78.437	85.956
brushwood	27.868	43.589	59.765	34.304
vineyard	75.716	86.180	84.455	87.976
herbaceous	51.938	68.368	69.212	67.544
agricultural_land	57.725	73.197	68.50	78.585
plowed_land	40.911	58.066	60.814	55.556
swimming	37.081	54.101	73.596	42.771
snow	0.000	0.000	0.000	0.000
greenhouse	60.688	75.535	71.451	80.115
average	59.013	72.926	75.127	71.993

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

Normalized Confusion Matrix (precision)

Normalized Confusion Matrix (recall)

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

IGNF
/

FLAIR-INC_rgb_15cl_resnet34-unet