Upload EuroSAT.py with huggingface_hub

EuroSAT.py

@@ -0,0 +1,159 @@
+import os
+import datasets
+import numpy as np
+import tifffile
+import skimage.io as io
+import torch
+
+from typing import ClassVar
+
+class EuroSATConfig(datasets.BuilderConfig):
+    all_band_names = (
+        'B01',
+        'B02',
+        'B03',
+        'B04',
+        'B05',
+        'B06',
+        'B07',
+        'B08',
+        'B8A',
+        'B09',
+        'B10',
+        'B11',
+        'B12',
+    )
+
+    rgb_bands = ('B04', 'B03', 'B02')
+
+    BAND_SETS: ClassVar[dict[str, tuple[str, ...]]] = {
+        'all': all_band_names,
+        'rgb': rgb_bands,
+    }
+    """BuilderConfig for EuroSAT"""
+    def __init__(self, bands=BAND_SETS['all'], **kwargs):
+        super(EuroSATConfig, self).__init__(**kwargs)
+        self.bands = bands
+        self.band_indices = [self.all_band_names.index(b) for b in bands if b in self.all_band_names].long()
+
+class EuroSAT(datasets.GeneratorBasedBuilder):
+    VERSION = datasets.Version("1.0.0")
+    
+    BUILDER_CONFIGS = [
+        EuroSATConfig(
+            name="default",
+            description="Default configuration"
+        ),
+    ]
+    
+    DEFAULT_CONFIG_NAME = "default"
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        if 'config' in kwargs:
+            user_config = kwargs['config']
+            if isinstance(user_config, dict):
+                for key, value in user_config.items():
+                    if hasattr(self.config, key):
+                        setattr(self.config, key, value)
+            else:
+                raise ValueError("check user_config, it should be a dict")
+        else:
+            raise ValueError("config is not specified in arguments")
+        
+        self.height = 64
+        self.width = 64
+        self.num_channels = self.config.band_indices.shape[0] # TODO: check out this
+        self.labels = [
+          "AnnualCrop",
+          "Forest", 
+          "HerbaceousVegetation",
+          "Highway",
+          "Industrial",
+          "Pasture",
+          "PermanentCrop",
+          "Residential",
+          "River", 
+          "SeaLake",
+        ]
+        # self.label_ids = {label: idx for idx, label in enumerate(self.labels)}
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            features=datasets.Features({
+                "image": datasets.Array3D(shape=(self.num_channels, self.height, self.width), dtype="float32"),
+                "label": datasets.ClassLabel(names=self.labels),
+            }),
+            supervised_keys=("image", "label")
+        )
+
+    def _split_generators(self, dl_manager):
+        data_dir = dl_manager.download_and_extract("yuxuanw8/EuroSAT") # TODO: check out the correct address for downloading and extracting
+
+        return [
+            datasets.SplitGenerator(
+                name="train",
+                gen_kwargs={
+                    "split_file": os.path.join(data_dir, "eurosat-train.txt"),
+                    "data_dir": os.path.joint(data_dir, "EuroSAT"), 
+                },
+            ),
+            datasets.SplitGenerator(
+                name="val",
+                gen_kwargs={
+                    "split_file": os.path.join(data_dir, "eurosat-val.txt"),
+                    "data_dir": os.path.joint(data_dir, "EuroSAT"),
+                },
+            ),
+            datasets.SplitGenerator(
+                name="test",
+                gen_kwargs={
+                    "split_file": os.path.join(data_dir, "eurosat-test.txt"),
+                    "data_dir": os.path.joint(data_dir, "EuroSAT"),
+                },
+            )
+        ]
+
+    def _generate_examples(self, split_file, data_dir):
+
+        """
+        split_file: the filename that lists all data points
+        data_dir: directory where the actual data is stored
+        """
+
+        """
+        data_dir should be in following structure:
+            - EuroSAT
+                - AnnualCrop
+                    - AnnualCrop_1.tif
+                    - AnnualCrop_2.tif
+                    - ...
+                - Forest
+                    - Forest_1.tif
+                    - FOrest_2.tif
+                    - ...
+                - ...
+        """
+
+        with open(split_file, 'r') as f:
+            for line in f:
+                line = line.strip()
+        
+                file_name = line.replace(".jpg", ".tif")
+                label = file_name.split('_')[0]
+                data_path = os.path.join(data_dir, label, file_name)
+                
+                img = tifffile.imread(data_path)
+
+                # permute img from HxWxC to CxHxW
+                np.transpose(img, (2, 0, 1))
+
+                # drop any channels if applicable
+                img = np.take(img, indices=self.config.band_indices, axis=0)
+
+                sample = {
+                    "image": img.astype(np.float32),
+                    "label": self.info.features['label'].str2int(label),
+                }
+
+                yield f"{label}_{file_name}", sample