Add conversion script

script.py

@@ -0,0 +1,50 @@
+from pathlib import Path
+
+import datasets
+import numpy as np
+from PIL import Image
+
+project_name = 'xiazeyu/WildfireSimMaps'
+
+map_names = sorted([x.name for x in Path('dataset').iterdir() if x.is_dir()])
+
+_CITATION = """\
+"""
+
+_DESCRIPTION = 'A real-world dataset for wildfire simulation.'
+
+_HOMEPAGE = 'https://huggingface.co/datasets/xiazeyu/WildfireSimMaps'
+
+_LICENSE = 'CC BY-NC 4.0'
+
+
+def load_map(map_name):
+    map_root = Path('dataset') / map_name
+
+    return {'canopy': np.array(Image.open(map_root / 'canopy.tif')),
+            'density': np.array(Image.open(map_root / 'density.tif')),
+            'slope': np.array(Image.open(map_root / 'slope.tif')), }
+
+
+data = {'name': [], 'canopy': [], 'density': [], "slope": [], 'shape': [], }
+
+for name in map_names:
+    map_data = load_map(name)
+    data['name'].append(name)
+    data['canopy'].append(map_data['canopy'].flatten())
+    data['density'].append(map_data['density'].flatten())
+    data['slope'].append(map_data['slope'].flatten())
+    data['shape'].append(map_data['canopy'].shape)
+
+features = datasets.Features({'name': datasets.Value('string'), 'canopy': datasets.Sequence(datasets.Value('int8')),
+                              'density': datasets.Sequence(datasets.Value('float32')),
+                              'slope': datasets.Sequence(datasets.Value('int8')),
+                              'shape': datasets.Sequence(datasets.Value('int16'), length=2), })
+data_info = datasets.DatasetInfo(description=_DESCRIPTION, features=features, homepage=_HOMEPAGE, license=_LICENSE,
+                                 citation=_CITATION, )
+
+ds = datasets.Dataset.from_dict(data, features=features, info=data_info, )
+
+ds.VERSION = datasets.Version("1.0.0")
+
+ds.push_to_hub(project_name)