Subsets in dataset iterators

Summary: For the new API, filtering iterators over sequences by subsets is quite helpful. The change is backwards compatible.

Reviewed By: bottler

Differential Revision: D42739669

fbshipit-source-id: d150a404aeaf42fd04a81304c63a4cba203f897d

Author: shapovalov

pytorch3d/implicitron/dataset/dataset_base.py

@@ -237,7 +237,7 @@ def __len__(self) -> int:
         raise NotImplementedError()
 
     def get_frame_numbers_and_timestamps(
-        self, idxs: Sequence[int]
+        self, idxs: Sequence[int], subset_filter: Optional[Sequence[str]] = None
     ) -> List[Tuple[int, float]]:
         """
         If the sequences in the dataset are videos rather than
@@ -251,7 +251,9 @@ def get_frame_numbers_and_timestamps(
         frames.
 
         Args:
-            idx: frame index in self
+            idxs: frame index in self
+            subset_filter: If given, an index in idxs is ignored if the
+                corresponding frame is not in any of the named subsets.
 
         Returns:
             tuple of
@@ -291,7 +293,7 @@ def category_to_sequence_names(self) -> Dict[str, List[str]]:
         return dict(c2seq)
 
     def sequence_frames_in_order(
-        self, seq_name: str
+        self, seq_name: str, subset_filter: Optional[Sequence[str]] = None
     ) -> Iterator[Tuple[float, int, int]]:
         """Returns an iterator over the frame indices in a given sequence.
         We attempt to first sort by timestamp (if they are available),
@@ -308,7 +310,9 @@ def sequence_frames_in_order(
         """
         # pyre-ignore[16]
         seq_frame_indices = self._seq_to_idx[seq_name]
-        nos_timestamps = self.get_frame_numbers_and_timestamps(seq_frame_indices)
+        nos_timestamps = self.get_frame_numbers_and_timestamps(
+            seq_frame_indices, subset_filter
+        )
 
         yield from sorted(
             [
@@ -317,11 +321,13 @@ def sequence_frames_in_order(
             ]
         )
 
-    def sequence_indices_in_order(self, seq_name: str) -> Iterator[int]:
+    def sequence_indices_in_order(
+        self, seq_name: str, subset_filter: Optional[Sequence[str]] = None
+    ) -> Iterator[int]:
         """Same as `sequence_frames_in_order` but returns the iterator over
         only dataset indices.
         """
-        for _, _, idx in self.sequence_frames_in_order(seq_name):
+        for _, _, idx in self.sequence_frames_in_order(seq_name, subset_filter):
             yield idx
 
     # frame_data_type is the actual type of frames returned by the dataset.

pytorch3d/implicitron/dataset/json_index_dataset.py

@@ -888,10 +888,16 @@ def _local_path(self, path: str) -> str:
         return self.path_manager.get_local_path(path)
 
     def get_frame_numbers_and_timestamps(
-        self, idxs: Sequence[int]
+        self, idxs: Sequence[int], subset_filter: Optional[Sequence[str]] = None
     ) -> List[Tuple[int, float]]:
         out: List[Tuple[int, float]] = []
         for idx in idxs:
+            if (
+                subset_filter is not None
+                and self.frame_annots[idx]["subset"] not in subset_filter
+            ):
+                continue
+
             # pyre-ignore[16]
             frame_annotation = self.frame_annots[idx]["frame_annotation"]
             out.append(

tests/implicitron/test_data_json_index.py

@@ -40,3 +40,41 @@ def test_loaders(self):
         self.assertEqual(len(data_sets.train), 81)
         self.assertEqual(len(data_sets.val), 102)
         self.assertEqual(len(data_sets.test), 102)
+
+    def test_visitor_subsets(self):
+        args = get_default_args(ImplicitronDataSource)
+        args.dataset_map_provider_class_type = "JsonIndexDatasetMapProvider"
+        dataset_args = args.dataset_map_provider_JsonIndexDatasetMapProvider_args
+        dataset_args.category = "skateboard"
+        dataset_args.dataset_root = "manifold://co3d/tree/extracted"
+        dataset_args.test_restrict_sequence_id = 0
+        dataset_args.dataset_JsonIndexDataset_args.limit_sequences_to = 1
+
+        data_source = ImplicitronDataSource(**args)
+        datasets, _ = data_source.get_datasets_and_dataloaders()
+        dataset = datasets.test
+
+        sequences = list(dataset.sequence_names())
+        self.assertEqual(len(sequences), 1)
+        i = 0
+        for seq in sequences:
+            last_ts = float("-Inf")
+            seq_frames = list(dataset.sequence_frames_in_order(seq))
+            self.assertEqual(len(seq_frames), 102)
+            for ts, _, idx in seq_frames:
+                self.assertEqual(i, idx)
+                i += 1
+                self.assertGreaterEqual(ts, last_ts)
+                last_ts = ts
+
+            last_ts = float("-Inf")
+            known_frames = list(dataset.sequence_frames_in_order(seq, "test_known"))
+            self.assertEqual(len(known_frames), 81)
+            for ts, _, _ in known_frames:
+                self.assertGreaterEqual(ts, last_ts)
+                last_ts = ts
+
+            known_indices = list(dataset.sequence_indices_in_order(seq, "test_known"))
+            self.assertEqual(len(known_indices), 81)
+
+            break  # testing only the first sequence