Datasets:

Formats:
csv
Languages:
English
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
arjunrao2000 commited on
Commit
c6016e9
·
verified ·
1 Parent(s): 47a52a3

First Update of metadata YAML

Browse files
Files changed (1) hide show
  1. README.md +79 -23
README.md CHANGED
@@ -1,42 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ## Geolayers-Data
2
- This dataset card contains usage instructions and metadata for all data-products released with our paper: 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.
3
 
4
- | **Dataset** | **Task Description** | **Multispectral Input** | **Model** | **Additional Data Layers** | **OOD Test Set Present?** |
5
- |-------------|----------------------|-------------------------|-----------|----------------------------|----------|
6
- | [SustainBench](https://arxiv.org/abs/2111.04724) | Farmland boundary delineation | Sentinel-2 RGB | U-Net | OSM rasters, EU-DEM | ✗ |
7
- | [EnviroAtlas](https://arxiv.org/abs/2202.14000) | Land-cover segmentation | NAIP RGB + NIR | FCN | [Prior](https://arxiv.org/abs/2202.14000), OSM rasters | |
8
- | [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 | ✓ |
9
- | [USAVars](https://arxiv.org/abs/2010.08168) | Tree-cover regression | NAIP RGB + NIR | ResNet-50 | OSM rasters, DEM | ✗ |
 
 
 
10
 
11
  ## 📦 Datasets & Georeferenced Auxiliary Layers
12
 
13
- ### SustainBench – Farmland Boundary Delineation
14
- * **Optical input:** Sentinel-2 RGB patches (224 × 224 px, 10 m GSD) covering French cropland in 2017; ≈ 1.6 k training images.
15
  * **Auxiliary layers (all geo-aligned):**
16
  * 19-channel OpenStreetMap (OSM) raster stack (roads, waterways, buildings, biome classes, …)
17
  * EU-DEM (20 m GSD, down-sampled to 10 m)
18
- * **Why:** OSM + DEM give an 8 % Dice boost when labels are scarce; gains appear once the training set drops below ≈ 700 images.
19
 
20
  ---
21
 
22
- ### EnviroAtlas – Land-Cover Segmentation
23
  * **Optical input:** NAIP 4-band RGB-NIR aerial imagery at 1 m resolution.
24
  * **Auxiliary layers:**
25
  * OSM rasters (roads, waterbodies, waterways)
26
  * **Prior** raster – a hand-crafted fusion of NLCD land-cover and OSM layers (PROC-STACK)
27
- * **Splits:** Train = Pittsburgh; OOD validation/test = Austin & Durham. Auxiliary layers raise OOD overall accuracy by ~4 pp without extra fine-tuning.
28
 
29
  ---
30
 
31
- ### BigEarthNet v2.0 – Multi-Label Land-Cover Classification
32
  * **Optical input:** 10-band Sentinel-2 tile pairs; ≈ 550 k patch/label pairs over 19 classes.
33
  * **Auxiliary layer:**
34
- * **SatCLIP location embedding** (256-D), one per image centre, injected as an extra ViT token (TOKEN-FUSE).
35
- * **Splits:** Grid-based; val/test tiles lie outside the training footprint (spatial OOD by design). SatCLIP token lifts macro-F1 by ~3 pp across *all* subset sizes.
36
 
37
  ---
38
 
39
- ### USAVars – Tree-Cover Regression
40
  * **Optical input:** NAIP RGB-NIR images (1 km² tiles); ≈ 100 k samples with tree-cover % labels.
41
  * **Auxiliary layers:**
42
  * Extended OSM raster stack (roads, buildings, land-use, biome classes, …)
@@ -44,10 +107,3 @@ This dataset card contains usage instructions and metadata for all data-products
44
  * **Notes:** Stacking the OSM raster boosts R² by 0.16 in the low-data regime (< 250 images); DEM is provided raw for flexibility.
45
 
46
  ---
47
- license: mit
48
- task_categories:
49
- - image-classification
50
- - image-segmentation
51
- tags:
52
- - climate
53
- ---
 
1
+ ---
2
+ # ======= 1) Basic info =======
3
+ pretty_name: "Geolayers"
4
+ language: en
5
+ language_creators:
6
+ - "found"
7
+ license: mit
8
+ multilinguality: monolingual
9
+ size_categories:
10
+ - 1K<n<100K
11
+ task_categories:
12
+ - image-classification
13
+ - image-segmentation
14
+
15
+ # ======= 2) How to cite =======
16
+ citation: |
17
+ @inproceedings{rao2025,
18
+ title={Using Multiple Input Modalities can Improve Data‐Efficiency and O.O.D. Generalization for ML with Satellite Imagery},
19
+ author={Arjun Rao and Esther Rolf},
20
+ year={2025},
21
+ booktitle={Under Review},
22
+ }
23
+
24
+ # ======= 3) Dataset structure =======
25
+ source_datasets:
26
+ - "SustainBench"
27
+ - "USAVars"
28
+ - "BigEarthNetv2.0"
29
+ - "EnviroAtlas"
30
+
31
+ # features:
32
+ # image:
33
+ # dtype: "uint8"
34
+ # shape: [3, 256, 256]
35
+ # osm_layers:
36
+ # dtype: "float32"
37
+ # shape: [4, 256, 256]
38
+ # label:
39
+ # ClassLabel:
40
+ # names: ["urban", "agriculture", "forest", "water", "bareground"]
41
+
42
+ # splits:
43
+ # train:
44
+ # name: "train"
45
+ # num_examples: 8000
46
+ # validation:
47
+ # name: "validation"
48
+ # num_examples: 1000
49
+ # test:
50
+ # name: "test"
51
+ # num_examples: 1000
52
+
53
+ # ======= 4) Other metadata =======
54
+ homepage: "https://huggingface.co/datasets/arjunrao2000/geolayers"
55
+ repository: "https://huggingface.co/datasets/arjunrao2000/geolayers"
56
+ download_size: 2.557e+10
57
+ tags:
58
+ - climate
59
+ ---
60
+
61
+
62
  ## Geolayers-Data
 
63
 
64
+ This dataset card contains usage instructions and metadata for all data-products released with our paper:
65
+ *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.
66
+
67
+ | **Dataset** | **Task Description** | **Multispectral Input** | **Model** | **Additional Data Layers** | **OOD Test Set Present?** |
68
+ |--------------------------------------|------------------------------------|-----------------------------|------------|-------------------------------------------------------|---------------------------|
69
+ | [SustainBench](https://arxiv.org/abs/2111.04724) | Farmland boundary delineation | Sentinel-2 RGB | U-Net | OSM rasters, EU-DEM | ✗ |
70
+ | [EnviroAtlas](https://arxiv.org/abs/2202.14000) | Land-cover segmentation | NAIP RGB + NIR | FCN | [Prior](https://arxiv.org/abs/2202.14000), OSM rasters | ✓ |
71
+ | [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 | ✓ |
72
+ | [USAVars](https://arxiv.org/abs/2010.08168) | Tree-cover regression | NAIP RGB + NIR | ResNet-50 | OSM rasters, DEM | ✗ |
73
 
74
  ## 📦 Datasets & Georeferenced Auxiliary Layers
75
 
76
+ ### SustainBench – Farmland Boundary Delineation
77
+ * **Optical input:** Sentinel-2 RGB patches (224×224 px, 10 m GSD) covering French cropland in 2017; ≈ 1.6 k training images.
78
  * **Auxiliary layers (all geo-aligned):**
79
  * 19-channel OpenStreetMap (OSM) raster stack (roads, waterways, buildings, biome classes, …)
80
  * EU-DEM (20 m GSD, down-sampled to 10 m)
81
+ * **Why:** OSM + DEM give an 8 % Dice boost when labels are scarce; gains appear once the training set drops below ≈ 700 images.
82
 
83
  ---
84
 
85
+ ### EnviroAtlas – Land-Cover Segmentation
86
  * **Optical input:** NAIP 4-band RGB-NIR aerial imagery at 1 m resolution.
87
  * **Auxiliary layers:**
88
  * OSM rasters (roads, waterbodies, waterways)
89
  * **Prior** raster – a hand-crafted fusion of NLCD land-cover and OSM layers (PROC-STACK)
90
+ * **Splits:** Train = Pittsburgh; OOD validation/test = Austin & Durham. Auxiliary layers raise OOD overall accuracy by ~4 pp without extra fine-tuning.
91
 
92
  ---
93
 
94
+ ### BigEarthNet v2.0 – Multi-Label Land-Cover Classification
95
  * **Optical input:** 10-band Sentinel-2 tile pairs; ≈ 550 k patch/label pairs over 19 classes.
96
  * **Auxiliary layer:**
97
+ * **SatCLIP** location embedding (256-D), one per image center, injected as an extra ViT token (TOKEN-FUSE).
98
+ * **Splits:** Grid-based; val/test tiles lie outside the training footprint (spatial OOD by design). SatCLIP token lifts macro-F1 by ~3 pp across *all* subset sizes.
99
 
100
  ---
101
 
102
+ ### USAVars – Tree-Cover Regression
103
  * **Optical input:** NAIP RGB-NIR images (1 km² tiles); ≈ 100 k samples with tree-cover % labels.
104
  * **Auxiliary layers:**
105
  * Extended OSM raster stack (roads, buildings, land-use, biome classes, …)
 
107
  * **Notes:** Stacking the OSM raster boosts R² by 0.16 in the low-data regime (< 250 images); DEM is provided raw for flexibility.
108
 
109
  ---