Update README.md
Browse files
README.md
CHANGED
|
@@ -143,13 +143,27 @@ cat bigearthnet.tar.gz.part-* \
|
|
| 143 |
|
| 144 |
Note that if this version of the dataset is used, SatCLIP embeddings would need to be re-computed on-the-fly. To use this dataset with the pre-computed SatCLIP embeddings, refer to the note below.
|
| 145 |
|
| 146 |
-
|
| 147 |
The second version of the BigEarthNetv2.0 dataset is stored in `data/bigearthnet/`. These datasets are stored as 3 H5PY datasets (`.h5`) for each split in the dataset.
|
| 148 |
This version of the processed dataset comes with (i) raw location co-ordinates, and (ii) pre-computed SatCLIP embeddings (L=10, ResNet50 image encoder backbone).
|
| 149 |
You may access these embeddings and location metadata with keys `location` and `satclip_embedding`.
|
| 150 |
|
| 151 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
|
|
|
|
| 153 |
### SustainBench – Farmland Boundary Delineation
|
| 154 |
* **Optical input:** Sentinel-2 RGB patches (224×224 px, 10 m GSD) covering French cropland in 2017; ≈ 1.6 k training images.
|
| 155 |
* **Auxiliary layers (all geo-aligned):**
|
|
|
|
| 143 |
|
| 144 |
Note that if this version of the dataset is used, SatCLIP embeddings would need to be re-computed on-the-fly. To use this dataset with the pre-computed SatCLIP embeddings, refer to the note below.
|
| 145 |
|
| 146 |
+
#### 💡 Do you want to try your own input fusion mechanism with BigEarthNetv2.0?
|
| 147 |
The second version of the BigEarthNetv2.0 dataset is stored in `data/bigearthnet/`. These datasets are stored as 3 H5PY datasets (`.h5`) for each split in the dataset.
|
| 148 |
This version of the processed dataset comes with (i) raw location co-ordinates, and (ii) pre-computed SatCLIP embeddings (L=10, ResNet50 image encoder backbone).
|
| 149 |
You may access these embeddings and location metadata with keys `location` and `satclip_embedding`.
|
| 150 |
|
| 151 |
+
### Usage Instructions for the SustainBench Farmland Boundary Delineation Dataset (Yeh et. al. (2021))
|
| 152 |
+
1. Unzip the archive in `data/sustainbench-field-boundary-delineation` with `unzip sustainbench.zip`
|
| 153 |
+
2. You should see a directory structure as follows:
|
| 154 |
+
```
|
| 155 |
+
dataset_release/
|
| 156 |
+
├── id_augmented_test_split_with_osm_new.h5.gz.zip
|
| 157 |
+
├── id_augmented_train_split_with_osm_new.h5.gz.zip
|
| 158 |
+
├── id_augmented_val_split_with_osm_new.h5.gz.zip
|
| 159 |
+
├── raw_id_augmented_test_split_with_osm_new.h5.gz.zip
|
| 160 |
+
├── raw_id_augmented_train_split_with_osm_new.h5.gz.zip
|
| 161 |
+
└── raw_id_augmented_val_split_with_osm_new.h5.gz.zip
|
| 162 |
+
```
|
| 163 |
+
3. Unzip all files using `unzip` and `pigz -d <path to .h5.gz file>`
|
| 164 |
+
There are two versions of data released: Datasets that begin with `id_augmented` refer to the version of the SustainBench farmland boundary delineation dataset with the OSM and DEM rasters pre-processed to RGB space following the application of the Gaussian Blur. Datasets that begin with `raw_id_augmented` contain the RGB imagery with 19 categorical rasters for OSM, and 1 raster for the DEM geographic input.
|
| 165 |
|
| 166 |
+
## 📦 <a name="geolayersused"></a> Datasets & Georeferenced Auxiliary Layers
|
| 167 |
### SustainBench – Farmland Boundary Delineation
|
| 168 |
* **Optical input:** Sentinel-2 RGB patches (224×224 px, 10 m GSD) covering French cropland in 2017; ≈ 1.6 k training images.
|
| 169 |
* **Auxiliary layers (all geo-aligned):**
|