FLAIR: French Land Cover from Aerospace Imagery
Land use and land cover classification from high-resolution aerial imagery. Annotators classify the primary land use category of aerial image patches and identify any secondary land uses present. Based on the FLAIR dataset from the French National Institute of Geographic and Forest Information (IGN).
Configuration Fileconfig.yaml
This Potato config reproduces the annotation task. Save it as config.yaml and run potato start config.yaml to try it.
# FLAIR: French Land Cover from Aerospace Imagery
# Based on Garioud et al., NeurIPS 2023
# Paper: https://arxiv.org/abs/2310.13336
# Dataset: https://ignf.github.io/FLAIR/
#
# Task: Land use and land cover classification from high-resolution aerial imagery
# Annotators classify the primary and secondary land use categories of
# aerial image patches at 0.2m spatial resolution from the French IGN survey.
#
# Annotation workflow:
# 1. Examine the aerial image patch
# 2. Classify the dominant/primary land use category
# 3. Select any secondary land uses also present in the patch
# 4. Rate the image quality
# 5. Add any additional observations
annotation_task_name: "FLAIR: French Land Cover from Aerospace Imagery"
task_dir: "."
data_files:
- sample-data.json
item_properties:
id_key: "id"
text_key: "image_url"
output_annotation_dir: "annotation_output/"
output_annotation_format: "json"
annotation_task_name: "FLAIR Land Cover Classification"
annotation_instructions: |
You are classifying land use and land cover from high-resolution (0.2m) aerial imagery collected over France.
**Instructions:**
1. Examine the aerial image patch carefully.
2. Select the **primary land use** category that best describes the dominant coverage in the image.
3. Select any **secondary land uses** that are also visible in the image (there may be multiple).
4. Rate the overall **image quality** for annotation purposes.
**Category Definitions:**
- **Building:** Residential, commercial, or industrial structures
- **Pervious surface:** Natural permeable surfaces (grass, soil in gardens/parks)
- **Impervious surface:** Artificial sealed surfaces (roads, parking lots, sidewalks)
- **Bare soil:** Exposed earth without vegetation or artificial covering
- **Water:** Rivers, lakes, ponds, canals, swimming pools
- **Coniferous:** Evergreen needle-leaved trees (pine, spruce, fir)
- **Deciduous:** Broadleaf trees that shed leaves seasonally (oak, beech, maple)
- **Brushwood:** Shrubs, bushes, and low woody vegetation
- **Vineyard:** Cultivated grapevines in rows
- **Herbaceous vegetation:** Grass, meadows, lawns, and non-woody plants
- **Agricultural land:** Active cropland with visible cultivation patterns
- **Plowed land:** Recently tilled fields without standing crops
- **Swimming pool:** Artificial water basins for recreation
- **Snow:** Snow-covered areas
- **Greenhouse:** Glass or plastic covered agricultural structures
annotation_schemes:
- annotation_type: radio
name: primary_land_use
description: "Select the dominant land use/land cover category in this image patch"
labels:
- name: "building"
tooltip: "Residential, commercial, or industrial structures"
- name: "pervious-surface"
tooltip: "Natural permeable surfaces such as grass or soil in gardens and parks"
- name: "impervious-surface"
tooltip: "Artificial sealed surfaces such as roads, parking lots, and sidewalks"
- name: "bare-soil"
tooltip: "Exposed earth without vegetation or artificial covering"
- name: "water"
tooltip: "Rivers, lakes, ponds, canals"
- name: "coniferous"
tooltip: "Evergreen needle-leaved trees (pine, spruce, fir)"
- name: "deciduous"
tooltip: "Broadleaf trees that shed leaves seasonally (oak, beech, maple)"
- name: "brushwood"
tooltip: "Shrubs, bushes, and low woody vegetation"
- name: "vineyard"
tooltip: "Cultivated grapevines arranged in rows"
- name: "herbaceous-vegetation"
tooltip: "Grass, meadows, lawns, and non-woody plants"
- name: "agricultural-land"
tooltip: "Active cropland with visible cultivation patterns"
- name: "plowed-land"
tooltip: "Recently tilled fields without standing crops"
- name: "swimming-pool"
tooltip: "Artificial water basins for recreation"
- name: "snow"
tooltip: "Snow-covered areas"
- name: "greenhouse"
tooltip: "Glass or plastic covered agricultural structures"
label_colors:
"building": "#e63946"
"pervious-surface": "#a8dadc"
"impervious-surface": "#6c757d"
"bare-soil": "#d4a373"
"water": "#1d3557"
"coniferous": "#2d6a4f"
"deciduous": "#52b788"
"brushwood": "#95d5b2"
"vineyard": "#7b2cbf"
"herbaceous-vegetation": "#b7e4c7"
"agricultural-land": "#ffd166"
"plowed-land": "#c9b99a"
"swimming-pool": "#00b4d8"
"snow": "#f8f9fa"
"greenhouse": "#e9ecef"
- annotation_type: multiselect
name: secondary_land_uses
description: "Select all other land use categories visible in this image (may select multiple)"
labels:
- name: "building"
tooltip: "Residential, commercial, or industrial structures"
- name: "pervious-surface"
tooltip: "Natural permeable surfaces"
- name: "impervious-surface"
tooltip: "Artificial sealed surfaces"
- name: "bare-soil"
tooltip: "Exposed earth"
- name: "water"
tooltip: "Water bodies"
- name: "coniferous"
tooltip: "Evergreen trees"
- name: "deciduous"
tooltip: "Broadleaf trees"
- name: "brushwood"
tooltip: "Shrubs and bushes"
- name: "vineyard"
tooltip: "Grapevines"
- name: "herbaceous-vegetation"
tooltip: "Grass and meadows"
- name: "agricultural-land"
tooltip: "Cropland"
- name: "plowed-land"
tooltip: "Tilled fields"
- name: "swimming-pool"
tooltip: "Pools"
- name: "snow"
tooltip: "Snow cover"
- name: "greenhouse"
tooltip: "Agricultural structures"
- annotation_type: radio
name: image_quality
description: "Rate the overall image quality for annotation purposes"
labels:
- name: "excellent"
tooltip: "Sharp, well-lit, no artifacts — ideal for annotation"
- name: "good"
tooltip: "Minor issues but land cover is clearly identifiable"
- name: "acceptable"
tooltip: "Some quality issues (slight blur, shadows) but annotation is feasible"
- name: "poor"
tooltip: "Significant quality issues that make annotation difficult"
- name: "unusable"
tooltip: "Image is too degraded for reliable annotation (heavy cloud, extreme blur)"
- annotation_type: text
name: notes
description: "Additional observations about the image patch (e.g., mixed-use areas, unusual features, boundary cases)"
html_layout: |
<div style="margin-bottom: 12px; padding: 10px; background: #e8f5e9; border-radius: 6px; border-left: 4px solid #2e7d32;">
<strong>Patch ID:</strong> {{patch_id}} |
<strong>Region:</strong> {{region}} |
<strong>Resolution:</strong> {{spatial_resolution}} |
<strong>Acquired:</strong> {{acquisition_date}}
</div>
<div style="text-align: center; margin-bottom: 12px;">
<img src="{{image_url}}" style="max-width: 100%; max-height: 600px; border: 2px solid #2e7d32; border-radius: 4px;" />
</div>
allow_all_users: true
instances_per_annotator: 50
annotation_per_instance: 2
allow_skip: true
skip_reason_required: false
Sample Datasample-data.json
[
{
"id": "flair_001",
"image_url": "https://example.com/flair/patches/D013_0001.png",
"patch_id": "D013_0001",
"region": "Ile-de-France",
"spatial_resolution": "0.2m",
"acquisition_date": "2020-06-15"
},
{
"id": "flair_002",
"image_url": "https://example.com/flair/patches/D033_0045.png",
"patch_id": "D033_0045",
"region": "Nouvelle-Aquitaine",
"spatial_resolution": "0.2m",
"acquisition_date": "2020-07-22"
}
]
// ... and 8 more itemsGet This Design
Clone or download from the repository
Quick start:
git clone https://github.com/davidjurgens/potato-showcase.git cd potato-showcase/image/flair-aerial-land-use potato start config.yaml
Dataset & paper
Garioud et al., NeurIPS 2023
Citation (BibTeX)
@inproceedings{garioud2023flair,
title = "{FLAIR}: {F}rench Land cover from {A}erospace {I}mage{R}y",
author = "Garioud, Anatol and Peillet, Stéphane and Bookjans, Eva and Giordano, Sébastien and Wattrelos, Boris",
booktitle = "Advances in Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track",
year = "2023",
url = "https://arxiv.org/abs/2310.13336"
}Details
Annotation Types
Domain
Use Cases
Tags
Found an issue or want to improve this design?
Open an IssueRelated Designs
CUB-200-2011: Fine-Grained Bird Species Dataset
CUB-200-2011 is a fine-grained bird image dataset of 11,788 photos across 200 species, each labeled with parts, attributes, and a bounding box (Wah et al., 2011). This Potato config reproduces the labeling task.
DOTA Aerial Image Object Detection
Oriented bounding box detection in aerial images (Xia et al., CVPR 2018). Detect 15 object categories with arbitrary orientations including planes, ships, vehicles, and sports facilities.
iWildCam Wildlife Detection & Classification
Camera trap image classification for wildlife monitoring (Beery et al., CVPR 2019). Classify wildlife species from camera trap images across diverse ecosystems worldwide.