Update README.md
Browse files
README.md
CHANGED
@@ -56,6 +56,8 @@ configs:
|
|
56 |
|
57 |
## Geolayers-Data
|
58 |
|
|
|
|
|
59 |
This dataset card contains usage instructions and metadata for all data-products released with our paper:
|
60 |
*Using Multiple Input Modalities can Improve Data-Efficiency and O.O.D. Generalization for ML with Satellite Imagery.* We release 3 modified versions of 3 benchmark datasets spanning land-cover segmentation, tree-cover regression, and multi-label land-cover classification tasks. These datasets are augmented with auxiliary, geographic inputs. A full list of contributed data products is shown in the table below.
|
61 |
|
@@ -64,7 +66,7 @@ This dataset card contains usage instructions and metadata for all data-products
|
|
64 |
| [SustainBench](https://arxiv.org/abs/2111.04724) | Farmland boundary delineation | Sentinel-2 RGB | U-Net | OSM rasters, EU-DEM | ✗ |
|
65 |
| [EnviroAtlas](https://arxiv.org/abs/2202.14000) | Land-cover segmentation | NAIP RGB + NIR | FCN | [Prior](https://arxiv.org/abs/2202.14000), OSM rasters | ✓ |
|
66 |
| [BigEarthNet v2.0](https://bigearth.net/static/documents/Description_BigEarthNet_v2.pdf) | Land-cover classification | Sentinel-2 (10 bands) | ViT | [SatCLIP](https://arxiv.org/abs/2311.17179) embeddings | ✓ |
|
67 |
-
| [USAVars](https://arxiv.org/abs/2010.08168) | Tree-cover regression | NAIP RGB + NIR | ResNet-50 | OSM rasters
|
68 |
|
69 |
## 📦 Datasets & Georeferenced Auxiliary Layers
|
70 |
|
@@ -98,7 +100,6 @@ This dataset card contains usage instructions and metadata for all data-products
|
|
98 |
* **Optical input:** NAIP RGB-NIR images (1 km² tiles); ≈ 100 k samples with tree-cover % labels.
|
99 |
* **Auxiliary layers:**
|
100 |
* Extended OSM raster stack (roads, buildings, land-use, biome classes, …)
|
101 |
-
* Continental Europe Digital Elevation Model (DEM) resampled to 10 m GSD
|
102 |
* **Notes:** Stacking the OSM raster boosts R² by 0.16 in the low-data regime (< 250 images); DEM is provided raw for flexibility.
|
103 |
|
104 |
## Usage Instructions
|
|
|
56 |
|
57 |
## Geolayers-Data
|
58 |
|
59 |
+

|
60 |
+
|
61 |
This dataset card contains usage instructions and metadata for all data-products released with our paper:
|
62 |
*Using Multiple Input Modalities can Improve Data-Efficiency and O.O.D. Generalization for ML with Satellite Imagery.* We release 3 modified versions of 3 benchmark datasets spanning land-cover segmentation, tree-cover regression, and multi-label land-cover classification tasks. These datasets are augmented with auxiliary, geographic inputs. A full list of contributed data products is shown in the table below.
|
63 |
|
|
|
66 |
| [SustainBench](https://arxiv.org/abs/2111.04724) | Farmland boundary delineation | Sentinel-2 RGB | U-Net | OSM rasters, EU-DEM | ✗ |
|
67 |
| [EnviroAtlas](https://arxiv.org/abs/2202.14000) | Land-cover segmentation | NAIP RGB + NIR | FCN | [Prior](https://arxiv.org/abs/2202.14000), OSM rasters | ✓ |
|
68 |
| [BigEarthNet v2.0](https://bigearth.net/static/documents/Description_BigEarthNet_v2.pdf) | Land-cover classification | Sentinel-2 (10 bands) | ViT | [SatCLIP](https://arxiv.org/abs/2311.17179) embeddings | ✓ |
|
69 |
+
| [USAVars](https://arxiv.org/abs/2010.08168) | Tree-cover regression | NAIP RGB + NIR | ResNet-50 | OSM rasters | ✗ |
|
70 |
|
71 |
## 📦 Datasets & Georeferenced Auxiliary Layers
|
72 |
|
|
|
100 |
* **Optical input:** NAIP RGB-NIR images (1 km² tiles); ≈ 100 k samples with tree-cover % labels.
|
101 |
* **Auxiliary layers:**
|
102 |
* Extended OSM raster stack (roads, buildings, land-use, biome classes, …)
|
|
|
103 |
* **Notes:** Stacking the OSM raster boosts R² by 0.16 in the low-data regime (< 250 images); DEM is provided raw for flexibility.
|
104 |
|
105 |
## Usage Instructions
|