Move `start_index` from `SampleFrames` to dataset level (#89)

configs/recognition/i3d/i3d_r50_video_32x2x1_100e_kinetics400_rgb.py

@@ -31,12 +31,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=32,
-        frame_interval=2,
-        num_clips=1,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
@@ -59,7 +54,6 @@ val_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=1,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -77,7 +71,6 @@ test_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/i3d/i3d_r50_video_inference_32x2x1_100e_kinetics400_rgb.py

@@ -30,7 +30,6 @@ test_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=1,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/r2plus1d/r2plus1d_r34_video_8x8x1_180e_kinetics400_rgb.py

@@ -38,12 +38,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=8,
-        frame_interval=8,
-        num_clips=1,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=8, frame_interval=8, num_clips=1),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
@@ -61,7 +56,6 @@ val_pipeline = [
         clip_len=8,
         frame_interval=8,
         num_clips=1,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -79,7 +73,6 @@ test_pipeline = [
         clip_len=8,
         frame_interval=8,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/r2plus1d/r2plus1d_r34_video_inference_8x8x1_180e_kinetics400_rgb.py

@@ -38,7 +38,6 @@ test_pipeline = [
         clip_len=8,
         frame_interval=8,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/slowfast/slowfast_r50_video_4x16x1_256e_kinetics400_rgb.py

@@ -45,12 +45,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=32,
-        frame_interval=2,
-        num_clips=1,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
@@ -68,7 +63,6 @@ val_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=1,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -86,7 +80,6 @@ test_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/slowfast/slowfast_r50_video_inference_4x16x1_256e_kinetics400_rgb.py

@@ -48,7 +48,6 @@ test_pipeline = [
         clip_len=32,
         frame_interval=2,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/slowonly/slowonly_r50_video_4x16x1_256e_kinetics400_rgb.py

@@ -28,12 +28,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=4,
-        frame_interval=16,
-        num_clips=1,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=4, frame_interval=16, num_clips=1),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
@@ -51,7 +46,6 @@ val_pipeline = [
         clip_len=4,
         frame_interval=16,
         num_clips=1,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -69,7 +63,6 @@ test_pipeline = [
         clip_len=4,
         frame_interval=16,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/slowonly/slowonly_r50_video_inference_4x16x1_256e_kinetics400_rgb.py

@@ -28,7 +28,6 @@ test_pipeline = [
         clip_len=4,
         frame_interval=16,
         num_clips=10,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/tsm/tsm_r50_video_1x1x8_50e_kinetics400_rgb.py

@@ -30,12 +30,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=1,
-        frame_interval=1,
-        num_clips=8,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
@@ -59,7 +54,6 @@ val_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -77,7 +71,6 @@ test_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/tsm/tsm_r50_video_inference_1x1x8_100e_kinetics400_rgb.py

@@ -30,7 +30,6 @@ test_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/tsn/tsn_r50_video_1x1x8_100e_kinetics400_rgb.py

@@ -28,12 +28,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='SampleFrames',
-        clip_len=1,
-        frame_interval=1,
-        num_clips=8,
-        start_index=0),
+    dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
     dict(type='DecordDecode'),
     dict(
         type='MultiScaleCrop',
@@ -55,7 +50,6 @@ val_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -73,7 +67,6 @@ test_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=25,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/tsn/tsn_r50_video_dense_1x1x8_100e_kinetics400_rgb.py

@@ -28,12 +28,7 @@ img_norm_cfg = dict(
     mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
 train_pipeline = [
     dict(type='DecordInit'),
-    dict(
-        type='DenseSampleFrames',
-        clip_len=1,
-        frame_interval=1,
-        num_clips=8,
-        start_index=0),
+    dict(type='DenseSampleFrames', clip_len=1, frame_interval=1, num_clips=8),
     dict(type='DecordDecode'),
     dict(
         type='MultiScaleCrop',
@@ -55,7 +50,6 @@ val_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),
@@ -73,7 +67,6 @@ test_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=8,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

configs/recognition/tsn/tsn_r50_video_inference_1x1x3_100e_kinetics400_rgb.py

@@ -27,7 +27,6 @@ test_pipeline = [
         clip_len=1,
         frame_interval=1,
         num_clips=25,
-        start_index=0,
         test_mode=True),
     dict(type='DecordDecode'),
     dict(type='Resize', scale=(-1, 256)),

mmaction/apis/inference.py

@@ -88,15 +88,22 @@ def inference_recognizer(model, video_path, label_path, use_frames=False):
     if use_frames:
         filename_tmpl = cfg.data.test.get('filename_tmpl', 'img_{:05}.jpg')
         modality = cfg.data.test.get('modality', 'RGB')
+        start_index = cfg.data.test.get('start_index', 1)
         data = dict(
             frame_dir=video_path,
             total_frames=len(os.listdir(video_path)),
             # assuming files in ``video_path`` are all named with ``filename_tmpl``  # noqa: E501
             label=-1,
+            start_index=start_index,
             filename_tmpl=filename_tmpl,
             modality=modality)
     else:
-        data = dict(filename=video_path, label=-1, modality='RGB')
+        start_index = cfg.data.test.get('start_index', 0)
+        data = dict(
+            filename=video_path,
+            label=-1,
+            start_index=start_index,
+            modality='RGB')
     data = test_pipeline(data)
     data = collate([data], samples_per_gpu=1)
     if next(model.parameters()).is_cuda:

mmaction/datasets/base.py

@@ -32,6 +32,10 @@ class BaseDataset(Dataset, metaclass=ABCMeta):
             dataset. Default: False.
         num_classes (int): Number of classes of the dataset, used in
             multi-class datasets. Default: None.
+        start_index (int): Specify a start index for frames in consideration of
+            different filename format. However, when taking videos as input,
+            it should be set to 0, since frames loaded from videos count
+            from 0. Default: 1.
         modality (str): Modality of data. Support 'RGB', 'Flow'.
             Default: 'RGB'.
     """
@@ -43,6 +47,7 @@ class BaseDataset(Dataset, metaclass=ABCMeta):
                  test_mode=False,
                  multi_class=False,
                  num_classes=None,
+                 start_index=1,
                  modality='RGB'):
         super().__init__()
 
@@ -52,6 +57,7 @@ class BaseDataset(Dataset, metaclass=ABCMeta):
         self.test_mode = test_mode
         self.multi_class = multi_class
         self.num_classes = num_classes
+        self.start_index = start_index
         self.modality = modality
         self.pipeline = Compose(pipeline)
         self.video_infos = self.load_annotations()
@@ -83,12 +89,14 @@ class BaseDataset(Dataset, metaclass=ABCMeta):
         """Prepare the frames for training given the index."""
         results = copy.deepcopy(self.video_infos[idx])
         results['modality'] = self.modality
+        results['start_index'] = self.start_index
         return self.pipeline(results)
 
     def prepare_test_frames(self, idx):
         """Prepare the frames for testing given the index."""
         results = copy.deepcopy(self.video_infos[idx])
         results['modality'] = self.modality
+        results['start_index'] = self.start_index
         return self.pipeline(results)
 
     def __len__(self):

mmaction/datasets/pipelines/loading.py

@@ -17,18 +17,14 @@ from ..registry import PIPELINES
 class SampleFrames(object):
     """Sample frames from the video.
 
-    Required keys are "filename", "total_frames", added or modified keys are
-    "frame_inds", "frame_interval" and "num_clips".
+    Required keys are "filename", "total_frames", "start_index" , added or
+    modified keys are "frame_inds", "frame_interval" and "num_clips".
 
     Args:
         clip_len (int): Frames of each sampled output clip.
         frame_interval (int): Temporal interval of adjacent sampled frames.
             Default: 1.
         num_clips (int): Number of clips to be sampled. Default: 1.
-        start_index (int): Specify a start index for frames in consideration of
-            different filename format. However, when taking videos as input,
-            it should be set to 0, since frames loaded from videos count
-            from 0. Default: 1.
         temporal_jitter (bool): Whether to apply temporal jittering.
             Default: False.
         twice_sample (bool): Whether to use twice sample when testing.
@@ -39,28 +35,35 @@ class SampleFrames(object):
             Default: 'loop'.
         test_mode (bool): Store True when building test or validation dataset.
             Default: False.
+        start_index (None): This argument is deprecated and moved to dataset
+            class (``BaseDataset``, ``VideoDatset``, ``RawframeDataset``, etc),
+            see this: https://github.com/open-mmlab/mmaction2/pull/89.
     """
 
     def __init__(self,
                  clip_len,
                  frame_interval=1,
                  num_clips=1,
-                 start_index=1,
                  temporal_jitter=False,
                  twice_sample=False,
                  out_of_bound_opt='loop',
-                 test_mode=False):
+                 test_mode=False,
+                 start_index=None):
 
         self.clip_len = clip_len
         self.frame_interval = frame_interval
         self.num_clips = num_clips
-        self.start_index = start_index
         self.temporal_jitter = temporal_jitter
         self.twice_sample = twice_sample
         self.out_of_bound_opt = out_of_bound_opt
         self.test_mode = test_mode
         assert self.out_of_bound_opt in ['loop', 'repeat_last']
 
+        if start_index is not None:
+            warnings.warn('No longer support "start_index" in "SampleFrames", '
+                          'it should be set in dataset class, see this pr: '
+                          'https://github.com/open-mmlab/mmaction2/pull/89')
+
     def _get_train_clips(self, num_frames):
         """Get clip offsets in train mode.
 
@@ -165,7 +168,9 @@ class SampleFrames(object):
             frame_inds = new_inds
         else:
             raise ValueError('Illegal out_of_bound option.')
-        frame_inds = np.concatenate(frame_inds) + self.start_index
+
+        start_index = results['start_index']
+        frame_inds = np.concatenate(frame_inds) + start_index
         results['frame_inds'] = frame_inds.astype(np.int)
         results['clip_len'] = self.clip_len
         results['frame_interval'] = self.frame_interval
@@ -185,8 +190,6 @@ class DenseSampleFrames(SampleFrames):
         frame_interval (int): Temporal interval of adjacent sampled frames.
             Default: 1.
         num_clips (int): Number of clips to be sampled. Default: 1.
-        start_index (int): Specify a start index for frames in consideration of
-            different filename format. Default: 1.
         sample_range (int): Total sample range for dense sample.
             Default: 64.
         num_sample_positions (int): Number of sample start positions, Which is
@@ -201,7 +204,6 @@ class DenseSampleFrames(SampleFrames):
                  clip_len,
                  frame_interval=1,
                  num_clips=1,
-                 start_index=1,
                  sample_range=64,
                  num_sample_positions=10,
                  temporal_jitter=False,
@@ -211,7 +213,6 @@ class DenseSampleFrames(SampleFrames):
             clip_len,
             frame_interval,
             num_clips,
-            start_index,
             temporal_jitter,
             out_of_bound_opt=out_of_bound_opt,
             test_mode=test_mode)
@@ -285,10 +286,6 @@ class SampleProposalFrames(SampleFrames):
             Default: 1.
         test_interval (int): Temporal interval of adjacent sampled frames
             in test mode. Default: 6.
-        start_index (int): Specify a start index for frames in consideration of
-            different filename format. However, when taking videos as input,
-            it should be set to 0, since frames loaded from videos count
-            from 0. Default: 1.
         temporal_jitter (bool): Whether to apply temporal jittering.
             Default: False.
         mode (str): Choose 'train', 'val' or 'test' mode.
@@ -302,13 +299,11 @@ class SampleProposalFrames(SampleFrames):
                  aug_ratio,
                  frame_interval=1,
                  test_interval=6,
-                 start_index=1,
                  temporal_jitter=False,
                  mode='train'):
         super().__init__(
             clip_len,
             frame_interval=frame_interval,
-            start_index=start_index,
             temporal_jitter=temporal_jitter)
         self.body_segments = body_segments
         self.aug_segments = aug_segments
@@ -500,7 +495,8 @@ class SampleProposalFrames(SampleFrames):
                 self.frame_interval, size=len(frame_inds))
             frame_inds += perframe_offsets
 
-        frame_inds = np.mod(frame_inds, total_frames) + self.start_index
+        start_index = results['start_index']
+        frame_inds = np.mod(frame_inds, total_frames) + start_index
 
         results['frame_inds'] = np.array(frame_inds).astype(np.int)
         results['clip_len'] = self.clip_len

mmaction/datasets/rawframe_dataset.py

@@ -85,11 +85,12 @@ class RawframeDataset(BaseDataset):
                  with_offset=False,
                  multi_class=False,
                  num_classes=None,
+                 start_index=1,
                  modality='RGB'):
         self.filename_tmpl = filename_tmpl
         self.with_offset = with_offset
         super().__init__(ann_file, pipeline, data_prefix, test_mode,
-                         multi_class, num_classes, modality)
+                         multi_class, num_classes, start_index, modality)
 
     def load_annotations(self):
         """Load annotation file to get video information."""
@@ -134,6 +135,7 @@ class RawframeDataset(BaseDataset):
         results = copy.deepcopy(self.video_infos[idx])
         results['filename_tmpl'] = self.filename_tmpl
         results['modality'] = self.modality
+        results['start_index'] = self.start_index
         return self.pipeline(results)
 
     def prepare_test_frames(self, idx):
@@ -141,6 +143,7 @@ class RawframeDataset(BaseDataset):
         results = copy.deepcopy(self.video_infos[idx])
         results['filename_tmpl'] = self.filename_tmpl
         results['modality'] = self.modality
+        results['start_index'] = self.start_index
         return self.pipeline(results)
 
     def evaluate(self,

mmaction/datasets/video_dataset.py

@@ -27,8 +27,21 @@ class VideoDataset(BaseDataset):
         some/path/003.mp4 2
         some/path/004.mp4 3
         some/path/005.mp4 3
+
+
+    Args:
+        ann_file (str): Path to the annotation file.
+        pipeline (list[dict | callable]): A sequence of data transforms.
+        start_index (int): Specify a start index for frames in consideration of
+            different filename format. However, when taking videos as input,
+            it should be set to 0, since frames loaded from videos count
+            from 0. Default: 0.
+        **kwargs: Keyword arguments for ``BaseDataset``.
     """
 
+    def __init__(self, ann_file, pipeline, start_index=0, **kwargs):
+        super().__init__(ann_file, pipeline, start_index=start_index, **kwargs)
+
     def load_annotations(self):
         """Load annotation file to get video information."""
         video_infos = []

tests/test_data/test_dataset.py

@@ -59,6 +59,7 @@ class TestDataset(object):
         assert rawframe_infos == [
             dict(frame_dir=frame_dir, total_frames=5, label=127)
         ] * 2
+        assert rawframe_dataset.start_index == 1
 
     def test_rawframe_dataset_with_offset(self):
         rawframe_dataset = RawframeDataset(
@@ -71,6 +72,7 @@ class TestDataset(object):
         assert rawframe_infos == [
             dict(frame_dir=frame_dir, offset=2, total_frames=5, label=127)
         ] * 2
+        assert rawframe_dataset.start_index == 1
 
     def test_rawframe_dataset_multi_label(self):
         rawframe_dataset = RawframeDataset(
@@ -90,6 +92,7 @@ class TestDataset(object):
             assert info['frame_dir'] == frame_dir
             assert info['total_frames'] == 5
             assert torch.all(info['label'] == label)
+        assert rawframe_dataset.start_index == 1
 
     def test_dataset_realpath(self):
         dataset = RawframeDataset(self.frame_ann_file, self.frame_pipeline,
@@ -100,14 +103,20 @@ class TestDataset(object):
         assert dataset.data_prefix == 's3://good'
 
     def test_video_dataset(self):
-        video_dataset = VideoDataset(self.video_ann_file, self.video_pipeline,
-                                     self.data_prefix)
+        video_dataset = VideoDataset(
+            self.video_ann_file,
+            self.video_pipeline,
+            data_prefix=self.data_prefix)
         video_infos = video_dataset.video_infos
         video_filename = osp.join(self.data_prefix, 'test.mp4')
         assert video_infos == [dict(filename=video_filename, label=0)] * 2
+        assert video_dataset.start_index == 0
 
     def test_rawframe_pipeline(self):
-        target_keys = ['frame_dir', 'total_frames', 'label', 'filename_tmpl']
+        target_keys = [
+            'frame_dir', 'total_frames', 'label', 'filename_tmpl',
+            'start_index', 'modality'
+        ]
 
         # RawframeDataset not in test mode
         rawframe_dataset = RawframeDataset(
@@ -129,6 +138,17 @@ class TestDataset(object):
         result = rawframe_dataset[0]
         assert self.check_keys_contain(result.keys(), target_keys)
 
+        # RawframeDataset with offset
+        rawframe_dataset = RawframeDataset(
+            self.frame_ann_file_with_offset,
+            self.frame_pipeline,
+            self.data_prefix,
+            with_offset=True,
+            num_classes=400,
+            test_mode=False)
+        result = rawframe_dataset[0]
+        assert self.check_keys_contain(result.keys(), target_keys + ['offset'])
+
         # RawframeDataset in test mode
         rawframe_dataset = RawframeDataset(
             self.frame_ann_file,
@@ -149,14 +169,25 @@ class TestDataset(object):
         result = rawframe_dataset[0]
         assert self.check_keys_contain(result.keys(), target_keys)
 
+        # RawframeDataset with offset
+        rawframe_dataset = RawframeDataset(
+            self.frame_ann_file_with_offset,
+            self.frame_pipeline,
+            self.data_prefix,
+            with_offset=True,
+            num_classes=400,
+            test_mode=True)
+        result = rawframe_dataset[0]
+        assert self.check_keys_contain(result.keys(), target_keys + ['offset'])
+
     def test_video_pipeline(self):
-        target_keys = ['filename', 'label']
+        target_keys = ['filename', 'label', 'start_index', 'modality']
 
         # VideoDataset not in test mode
         video_dataset = VideoDataset(
             self.video_ann_file,
             self.video_pipeline,
-            self.data_prefix,
+            data_prefix=self.data_prefix,
             test_mode=False)
         result = video_dataset[0]
         assert self.check_keys_contain(result.keys(), target_keys)
@@ -165,7 +196,7 @@ class TestDataset(object):
         video_dataset = VideoDataset(
             self.video_ann_file,
             self.video_pipeline,
-            self.data_prefix,
+            data_prefix=self.data_prefix,
             test_mode=True)
         result = video_dataset[0]
         assert self.check_keys_contain(result.keys(), target_keys)
@@ -221,8 +252,10 @@ class TestDataset(object):
             ['top1_acc', 'top5_acc', 'mean_class_accuracy'])
 
     def test_video_evaluate(self):
-        video_dataset = VideoDataset(self.video_ann_file, self.video_pipeline,
-                                     self.data_prefix)
+        video_dataset = VideoDataset(
+            self.video_ann_file,
+            self.video_pipeline,
+            data_prefix=self.data_prefix)
 
         with pytest.raises(TypeError):
             # results must be a list
@@ -248,10 +281,14 @@ class TestDataset(object):
             ['top1_acc', 'top5_acc', 'mean_class_accuracy'])
 
     def test_base_dataset(self):
-        video_dataset = VideoDataset(self.video_ann_file, self.video_pipeline,
-                                     self.data_prefix)
+        video_dataset = VideoDataset(
+            self.video_ann_file,
+            self.video_pipeline,
+            data_prefix=self.data_prefix,
+            start_index=3)
         assert len(video_dataset) == 2
         assert type(video_dataset[0]) == dict
+        assert video_dataset.start_index == 3
 
     def test_repeat_dataset(self):
         rawframe_dataset = RawframeDataset(self.frame_ann_file,

tests/test_data/test_loading.py

@@ -60,11 +60,15 @@ class TestLoading(object):
         cls.flow_filename_tmpl = '{}_{:05d}.jpg'
         video_total_frames = len(mmcv.VideoReader(cls.video_path))
         cls.video_results = dict(
-            filename=cls.video_path, label=1, total_frames=video_total_frames)
+            filename=cls.video_path,
+            label=1,
+            total_frames=video_total_frames,
+            start_index=0)
         cls.frame_results = dict(
             frame_dir=cls.img_dir,
             total_frames=cls.total_frames,
             filename_tmpl=cls.filename_tmpl,
+            start_index=1,
             modality='RGB',
             offset=0,
             label=1)
@@ -92,6 +96,7 @@ class TestLoading(object):
             video_id='test_imgs',
             total_frames=cls.total_frames,
             filename_tmpl=cls.filename_tmpl,
+            start_index=1,
             out_props=[[['test_imgs',
                          ExampleSSNInstance(1, 4, 10, 1, 1, 1)], 0],
                        [['test_imgs',
@@ -103,6 +108,12 @@ class TestLoading(object):
             'total_frames'
         ]
 
+        with pytest.warns(UserWarning):
+            # start_index has been deprecated
+            config = dict(
+                clip_len=3, frame_interval=1, num_clips=5, start_index=1)
+            SampleFrames(**config)
+
         # Sample Frame with no temporal_jitter
         # clip_len=3, frame_interval=1, num_clips=5
         video_result = copy.deepcopy(self.video_results)
@@ -116,6 +127,8 @@ class TestLoading(object):
         assert len(sample_frames_results['frame_inds']) == 15
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 15
+        assert np.max(sample_frames_results['frame_inds']) <= 5
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with no temporal_jitter
         # clip_len=5, frame_interval=1, num_clips=5,
@@ -150,6 +163,8 @@ class TestLoading(object):
         frame_inds = sample_frames_results['frame_inds'].reshape([5, 5])
         for i in range(5):
             assert check_monotonous(frame_inds[i])
+        assert np.max(sample_frames_results['frame_inds']) <= 5
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with temporal_jitter
         # clip_len=4, frame_interval=2, num_clips=5
@@ -164,6 +179,8 @@ class TestLoading(object):
         assert len(sample_frames_results['frame_inds']) == 20
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 20
+        assert np.max(sample_frames_results['frame_inds']) <= 5
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with no temporal_jitter in test mode
         # clip_len=4, frame_interval=1, num_clips=6
@@ -182,6 +199,8 @@ class TestLoading(object):
         assert len(sample_frames_results['frame_inds']) == 24
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 24
+        assert np.max(sample_frames_results['frame_inds']) <= 5
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with no temporal_jitter in test mode
         # clip_len=3, frame_interval=1, num_clips=6
@@ -200,6 +219,8 @@ class TestLoading(object):
         assert len(sample_frames_results['frame_inds']) == 18
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 18
+        assert np.max(sample_frames_results['frame_inds']) <= 5
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with no temporal_jitter to get clip_offsets
         # clip_len=1, frame_interval=1, num_clips=8
@@ -223,7 +244,7 @@ class TestLoading(object):
                            np.array([1, 2, 2, 3, 4, 5, 5, 6]))
 
         # Sample Frame with no temporal_jitter to get clip_offsets
-        # clip_len=1, frame_interval=1, num_clips=8, start_index=0
+        # clip_len=1, frame_interval=1, num_clips=8
         video_result = copy.deepcopy(self.video_results)
         frame_result = copy.deepcopy(self.frame_results)
         frame_result['total_frames'] = 6
@@ -231,18 +252,18 @@ class TestLoading(object):
             clip_len=1,
             frame_interval=1,
             num_clips=8,
-            start_index=0,
             temporal_jitter=False,
             test_mode=True)
         sample_frames = SampleFrames(**config)
         sample_frames_results = sample_frames(video_result)
+        assert sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
         assert len(sample_frames_results['frame_inds']) == 8
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 8
         assert_array_equal(sample_frames_results['frame_inds'],
-                           np.array([0, 1, 1, 2, 3, 4, 4, 5]))
+                           np.array([1, 2, 2, 3, 4, 5, 5, 6]))
 
         # Sample Frame with no temporal_jitter to get clip_offsets zero
         # clip_len=6, frame_interval=1, num_clips=1
@@ -257,6 +278,7 @@ class TestLoading(object):
             test_mode=True)
         sample_frames = SampleFrames(**config)
         sample_frames_results = sample_frames(video_result)
+        assert sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
         assert len(sample_frames_results['frame_inds']) == 6
@@ -278,11 +300,14 @@ class TestLoading(object):
             test_mode=False)
         sample_frames = SampleFrames(**config)
         sample_frames_results = sample_frames(video_result)
+        assert sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
         assert len(sample_frames_results['frame_inds']) == 240
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 240
+        assert np.max(sample_frames_results['frame_inds']) <= 30
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame with no temporal_jitter to get clip_offsets
         # clip_len=1, frame_interval=1, num_clips=8
@@ -299,6 +324,7 @@ class TestLoading(object):
         sample_frames_results = sample_frames(video_result)
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
+        assert sample_frames_results['start_index'] == 0
         assert len(sample_frames_results['frame_inds']) == 8
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 8
@@ -318,11 +344,14 @@ class TestLoading(object):
             test_mode=False)
         sample_frames = SampleFrames(**config)
         sample_frames_results = sample_frames(video_result)
+        assert sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
         assert len(sample_frames_results['frame_inds']) == 24
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 24
+        assert np.max(sample_frames_results['frame_inds']) <= 10
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
         # Sample Frame using twice sample
         # clip_len=12, frame_interval=1, num_clips=2
@@ -338,11 +367,14 @@ class TestLoading(object):
             test_mode=True)
         sample_frames = SampleFrames(**config)
         sample_frames_results = sample_frames(video_result)
+        assert sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(sample_frames_results.keys(),
                                        target_keys)
         assert len(sample_frames_results['frame_inds']) == 48
         sample_frames_results = sample_frames(frame_result)
         assert len(sample_frames_results['frame_inds']) == 48
+        assert np.max(sample_frames_results['frame_inds']) <= 40
+        assert np.min(sample_frames_results['frame_inds']) >= 1
 
     def test_dense_sample_frames(self):
         target_keys = [
@@ -362,6 +394,7 @@ class TestLoading(object):
             test_mode=True)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 240
@@ -376,6 +409,7 @@ class TestLoading(object):
             clip_len=4, frame_interval=1, num_clips=6, temporal_jitter=False)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 24
@@ -395,6 +429,7 @@ class TestLoading(object):
             test_mode=True)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 240
@@ -413,6 +448,7 @@ class TestLoading(object):
             temporal_jitter=False)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 24
@@ -431,6 +467,7 @@ class TestLoading(object):
             temporal_jitter=False)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 24
@@ -452,6 +489,7 @@ class TestLoading(object):
             test_mode=True)
         dense_sample_frames = DenseSampleFrames(**config)
         dense_sample_frames_results = dense_sample_frames(video_result)
+        assert dense_sample_frames_results['start_index'] == 0
         assert self.check_keys_contain(dense_sample_frames_results.keys(),
                                        target_keys)
         assert len(dense_sample_frames_results['frame_inds']) == 120
@@ -461,7 +499,7 @@ class TestLoading(object):
     def test_sample_proposal_frames(self):
         target_keys = [
             'frame_inds', 'clip_len', 'frame_interval', 'num_clips',
-            'total_frames'
+            'total_frames', 'start_index'
         ]
 
         # test error cases
@@ -475,7 +513,7 @@ class TestLoading(object):
                 aug_ratio=0.5,
                 temporal_jitter=False)
             sample_frames = SampleProposalFrames(**config)
-            sample_frames_results = sample_frames(proposal_result)
+            sample_frames(proposal_result)
 
         # test normal cases
         # Sample Frame with no temporal_jitter
@@ -839,7 +877,7 @@ class TestLoading(object):
     def test_rawframe_decode(self):
         target_keys = ['frame_inds', 'imgs', 'original_shape', 'modality']
 
-        # test frame selector with 2 dim input when start_index = 0
+        # test frame selector with 2 dim input
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(0, self.total_frames, 2)[:,
                                                                   np.newaxis]
@@ -887,7 +925,7 @@ class TestLoading(object):
                                              320, 3)
         assert results['original_shape'] == (240, 320)
 
-        # test frame selector with 1 dim input when start_index = 0
+        # test frame selector with 1 dim input
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(0, self.total_frames, 2)
         # since the test images start with index 1, we plus 1 to frame_inds
@@ -911,7 +949,6 @@ class TestLoading(object):
         assert results['original_shape'] == (240, 320)
 
         # test frame selector with 1 dim input for flow images
-        # when start_index = 0
         inputs = copy.deepcopy(self.flow_frame_results)
         inputs['frame_inds'] = np.arange(0, self.total_frames, 2)
         # since the test images start with index 1, we plus 1 to frame_inds