Create README.md

README.md

@@ -0,0 +1,69 @@
+# ScanObjectNN
+
+`scanobjectnn_PB_T50_RS_h5.zip` contains h5 files for the hard variant of the  ScanObjectNN benchmark.
+
+Dataset can be loaded as follows:
+
+```python
+import os.path as osp
+import os
+import torch
+import h5py
+import torch_geometric.transforms as T
+from torch_geometric.datasets import ModelNet
+from torch_geometric.data import InMemoryDataset, download_url, extract_zip, Data
+
+class ScanObjectNN(InMemoryDataset):
+    url = 'https://huggingface.co/datasets/cminst/ScanObjectNN/resolve/main/scanobjectnn_PB_T50_RS_h5.zip'
+
+    def __init__(self, root, train=True, transform=None, pre_transform=None, pre_filter=None):
+        self.train = train
+        super().__init__(root, transform, pre_transform, pre_filter)
+        path = self.processed_paths[0] if train else self.processed_paths[1]
+        self.load(path)
+
+    @property
+    def raw_file_names(self):
+        return [
+            osp.join('main_split', 'training_objectdataset_augmentedrot_scale75.h5'),
+            osp.join('main_split', 'test_objectdataset_augmentedrot_scale75.h5')
+        ]
+
+    @property
+    def processed_file_names(self):
+        return ['training.pt', 'test.pt']
+
+    def download(self):
+        path = download_url(self.url, self.raw_dir)
+        extract_zip(path, self.raw_dir)
+        os.unlink(path)
+
+    def process(self):
+        self.save(self.process_set('training'), self.processed_paths[0])
+        self.save(self.process_set('test'), self.processed_paths[1])
+
+    def process_set(self, split):
+        filename = f'{split}_objectdataset_augmentedrot_scale75.h5'
+
+        h5_path = osp.join(self.raw_dir, 'main_split', filename)
+
+        with h5py.File(h5_path, 'r') as f:
+            data = f['data'][:].astype('float32')
+            labels = f['label'][:].astype('int64')
+
+        data_list = []
+        for i in range(data.shape[0]):
+            pos = torch.from_numpy(data[i])
+            y = torch.tensor(labels[i]).view(1)
+
+            d = Data(pos=pos, y=y)
+            data_list.append(d)
+
+        if self.pre_filter is not None:
+            data_list = [d for d in data_list if self.pre_filter(d)]
+
+        if self.pre_transform is not None:
+            data_list = [self.pre_transform(d) for d in data_list]
+
+        return data_list
+```