Use `RawFrameDecode` to replace `FrameSelector` (#91)

configs/recognition/i3d/i3d_nl_r50_32x2x1_100e_kinetics400_rgb.py

@@ -37,7 +37,7 @@ 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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -59,7 +59,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -75,7 +75,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/i3d/i3d_r50_32x2x1_100e_kinetics400_rgb.py

@@ -31,7 +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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/i3d/i3d_r50_dense_32x2x1_100e_kinetics400_rgb.py

@@ -31,7 +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='DenseSampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/i3d/i3d_r50_fast_32x2x1_100e_kinetics400_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector', decoding_backend='turbojpeg'),
+    dict(type='RawFrameDecode', decoding_backend='turbojpeg'),
     dict(type='Resize', scale=(-1, 256), lazy=True),
     dict(
         type='MultiScaleCrop',
@@ -54,7 +54,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector', decoding_backend='turbojpeg'),
+    dict(type='RawFrameDecode', decoding_backend='turbojpeg'),
     dict(type='Resize', scale=(-1, 256), lazy=True),
     dict(type='CenterCrop', crop_size=224, lazy=True),
     dict(type='Flip', flip_ratio=0, lazy=True),
@@ -71,7 +71,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector', decoding_backend='turbojpeg'),
+    dict(type='RawFrameDecode', decoding_backend='turbojpeg'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/i3d/i3d_r50_heavy_8x8x1_100e_kinetics400_rgb.py

@@ -34,7 +34,7 @@ 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='SampleFrames', clip_len=8, frame_interval=8, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -56,7 +56,7 @@ val_pipeline = [
         frame_interval=8,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -72,7 +72,7 @@ test_pipeline = [
         frame_interval=8,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/r2plus1d/r2plus1d_r34_32x2x1_180e_kinetics400_rgb.py

@@ -37,7 +37,7 @@ 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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -54,7 +54,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -70,7 +70,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/r2plus1d/r2plus1d_r34_8x8x1_180e_kinetics400_rgb.py

@@ -37,7 +37,7 @@ 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='SampleFrames', clip_len=8, frame_interval=8, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -54,7 +54,7 @@ val_pipeline = [
         frame_interval=8,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -70,7 +70,7 @@ test_pipeline = [
         frame_interval=8,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowfast/slowfast_r50_4x16x1_256e_kinetics400_rgb.py

@@ -45,7 +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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -62,7 +62,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -78,7 +78,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowfast/slowfast_r50_8x8x1_256e_kinetics400_rgb.py

@@ -46,7 +46,7 @@ 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='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -63,7 +63,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -79,7 +79,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/data_benchmark/slowonly_r50_randomresizedcrop_320p_4x16x1_256e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=4, frame_interval=16, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=16,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=16,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/data_benchmark/slowonly_r50_randomresizedcrop_340x256_4x16x1_256e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=4, frame_interval=16, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=16,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=16,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/slowonly_r50_4x16x1_256e_kinetics400_flow.py

@@ -29,7 +29,7 @@ ann_file_test = 'data/kinetics400/kinetics_flow_val_list.txt'
 img_norm_cfg = dict(mean=[128, 128], std=[128, 128])
 train_pipeline = [
     dict(type='SampleFrames', clip_len=4, frame_interval=16, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -46,7 +46,7 @@ val_pipeline = [
         frame_interval=16,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),
@@ -62,7 +62,7 @@ test_pipeline = [
         frame_interval=16,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/slowonly_r50_4x16x1_256e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=4, frame_interval=16, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=16,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=16,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/slowonly_r50_8x8x1_256e_kinetics400_flow.py

@@ -29,7 +29,7 @@ ann_file_test = 'data/kinetics400/kinetics_flow_val_list.txt'
 img_norm_cfg = dict(mean=[128, 128], std=[128, 128])
 train_pipeline = [
     dict(type='SampleFrames', clip_len=8, frame_interval=8, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -46,7 +46,7 @@ val_pipeline = [
         frame_interval=8,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),
@@ -62,7 +62,7 @@ test_pipeline = [
         frame_interval=8,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/slowonly/slowonly_r50_8x8x1_256e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=8, frame_interval=8, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=8,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=8,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsm/tsm_nl_r50_1x1x8_50e_kinetics400_rgb.py

@@ -36,7 +36,7 @@ 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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -59,7 +59,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -75,7 +75,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsm/tsm_r101_1x1x8_50e_sthv2_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -52,7 +52,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),

configs/recognition/tsm/tsm_r50_1x1x16_50e_kinetics400_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=16,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=16,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsm/tsm_r50_1x1x16_50e_sthv2_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -52,7 +52,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=16,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=16,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),

configs/recognition/tsm/tsm_r50_1x1x8_50e_kinetics400_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsm/tsm_r50_1x1x8_50e_sthv1_rgb.py

@@ -30,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -52,7 +52,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),

configs/recognition/tsm/tsm_r50_dense_1x1x8_100e_kinetics400_rgb.py

@@ -30,7 +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='DenseSampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsm/tsm_temporal_pool_r50_1x1x8_50e_kinetics400_rgb.py

@@ -32,7 +32,7 @@ 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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -55,7 +55,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -71,7 +71,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_multiscalecrop_320p_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -50,7 +50,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -66,7 +66,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_multiscalecrop_340x256_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -50,7 +50,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -66,7 +66,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_randomresizedcrop_320p_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_randomresizedcrop_340x256_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_test_320p_1x1x25_10crop_100e_kinetics400_rgb.py

@@ -34,7 +34,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_test_320p_1x1x25_3crop_100e_kinetics400_rgb.py

@@ -34,7 +34,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_test_340x256_1x1x25_10crop_100e_kinetics400_rgb.py

@@ -34,7 +34,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/data_benchmark/tsn_r50_test_340x256_1x1x25_3crop_100e_kinetics400_rgb.py

@@ -30,7 +30,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r101_1x1x5_50e_mmit_rgb.py

@@ -31,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=5),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -53,7 +53,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=5,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -69,7 +69,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=5,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='MultiGroupCrop', crop_size=256, groups=1),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_1x1x16_50e_sthv1_rgb.py

@@ -29,7 +29,7 @@ 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='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -52,7 +52,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),

configs/recognition/tsn/tsn_r50_1x1x16_50e_sthv2_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -45,7 +45,7 @@ train_pipeline = [
 ]
 val_pipeline = [
     dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),
@@ -55,7 +55,7 @@ val_pipeline = [
 ]
 test_pipeline = [
     dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=16),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Normalize', **img_norm_cfg),

configs/recognition/tsn/tsn_r50_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -50,7 +50,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -66,7 +66,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_1x1x3_80e_ucf101_rgb.py

@@ -27,7 +27,7 @@ ann_file_test = 'data/ucf101/ucf101_val_split_{1,2,3}_rawframes.txt'
 img_norm_cfg = dict(mean=[104, 117, 128], std=[1, 1, 1], to_bgr=False)
 train_pipeline = [
     dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -49,7 +49,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -65,7 +65,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_1x1x6_100e_mit_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=6),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -50,7 +50,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=6,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -66,7 +66,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=6,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_1x1x8_50e_sthv1_rgb.py

@@ -29,7 +29,7 @@ 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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -51,7 +51,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_1x1x8_50e_sthv2_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -44,7 +44,7 @@ train_pipeline = [
 ]
 val_pipeline = [
     dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -54,7 +54,7 @@ val_pipeline = [
 ]
 test_pipeline = [
     dict(type='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_320p_1x1x3_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_320p_1x1x3_110e_kinetics400_flow.py

@@ -28,7 +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='SampleFrames', clip_len=5, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_320p_1x1x8_100e_kinetics400_rgb.py

@@ -28,7 +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='SampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_320p_1x1x8_110e_kinetics400_flow.py

@@ -28,7 +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='SampleFrames', clip_len=5, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='RandomResizedCrop'),
     dict(type='Resize', scale=(224, 224), keep_ratio=False),
@@ -45,7 +45,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -61,7 +61,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_dense_1x1x5_100e_kinetics400_rgb.py

@@ -28,7 +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='DenseSampleFrames', clip_len=1, frame_interval=1, num_clips=5),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -51,7 +51,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_dense_1x1x8_100e_kinetics400_rgb.py

@@ -28,7 +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='DenseSampleFrames', clip_len=1, frame_interval=1, num_clips=8),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -51,7 +51,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -67,7 +67,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=8,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),

configs/recognition/tsn/tsn_r50_inference_1x1x3_100e_kinetics400_rgb.py

@@ -27,7 +27,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

docs/config.md

@@ -236,8 +236,8 @@ train_pipeline = [  # List of training pipeline steps
         clip_len=1,  # Frames of each sampled output clip
         frame_interval=1,  # Temporal interval of adjacent sampled frames
         num_clips=3),  # Number of clips to be sampled
-    dict(  # Config of FrameSelector
-        type='FrameSelector'),  # Frame selector pipeline, selecting raw frames with given indices
+    dict(  # Config of RawFrameDecode
+        type='RawFrameDecode'),  # Load and decode Frames pipeline, picking raw frames with given indices
     dict(  # Config of Resize
         type='Resize',  # Resize pipeline
         scale=(-1, 256)),  # The scale to resize images
@@ -275,8 +275,8 @@ val_pipeline = [  # List of validation pipeline steps
         frame_interval=1,  # Temporal interval of adjacent sampled frames
         num_clips=3,  # Number of clips to be sampled
         test_mode=True),  # Whether to set test mode in sampling
-    dict(  # Config of FrameSelector
-        type='FrameSelector'),  # Frame selector pipeline, selecting raw frames with given indices
+    dict(  # Config of RawFrameDecode
+        type='RawFrameDecode'),  # Load and decode Frames pipeline, picking raw frames with given indices
     dict(  # Config of Resize
         type='Resize',  # Resize pipeline
         scale=(-1, 256)),  # The scale to resize images
@@ -307,8 +307,8 @@ test_pipeline = [  # List of testing pipeline steps
         frame_interval=1,  # Temporal interval of adjacent sampled frames
         num_clips=25,  # Number of clips to be sampled
         test_mode=True),  # Whether to set test mode in sampling
-    dict(  # Config of FrameSelector
-        type='FrameSelector'),  # Frame selector pipeline, selecting raw frames with given indices
+    dict(  # Config of RawFrameDecode
+        type='RawFrameDecode'),  # Load and decode Frames pipeline, picking raw frames with given indices
     dict(  # Config of Resize
         type='Resize',  # Resize pipeline
         scale=(-1, 256)),  # The scale to resize images
@@ -420,7 +420,7 @@ img_norm_cfg = dict(
 
 train_pipeline = [
     dict(type='SampleFrames', clip_len=32, frame_interval=2, num_clips=1),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -442,7 +442,7 @@ val_pipeline = [
         frame_interval=2,
         num_clips=1,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -458,7 +458,7 @@ test_pipeline = [
         frame_interval=2,
         num_clips=10,
         test_mode=True),
-    dict(type='FrameSelector'),
+    dict(type='RawFrameDecode'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='ThreeCrop', crop_size=256),
     dict(type='Flip', flip_ratio=0),

docs/tutorials/data_pipeline.md

@@ -25,7 +25,7 @@ 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='SampleFrames', clip_len=1, frame_interval=1, num_clips=3),
-    dict(type='FrameSelector', io_backend='disk'),
+    dict(type='RawFrameDecode', io_backend='disk'),
     dict(type='Resize', scale=(-1, 256)),
     dict(
         type='MultiScaleCrop',
@@ -47,7 +47,7 @@ val_pipeline = [
         frame_interval=1,
         num_clips=3,
         test_mode=True),
-    dict(type='FrameSelector', io_backend='disk'),
+    dict(type='RawFrameDecode', io_backend='disk'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='CenterCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -63,7 +63,7 @@ test_pipeline = [
         frame_interval=1,
         num_clips=25,
         test_mode=True),
-    dict(type='FrameSelector', io_backend='disk'),
+    dict(type='RawFrameDecode', io_backend='disk'),
     dict(type='Resize', scale=(-1, 256)),
     dict(type='TenCrop', crop_size=224),
     dict(type='Flip', flip_ratio=0),
@@ -96,7 +96,7 @@ For each operation, we list the related dict fields that are added/updated/remov
 - add: imgs, original_shape
 - update: *frame_inds
 
-`FrameSelector`
+`RawFrameDecode`
 - add: imgs, original_shape
 - update: *frame_inds
 
@@ -191,7 +191,7 @@ It is **noteworthy** that the first key, commonly `imgs`, will be used as the ma
          mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
     train_pipeline = [
         dict(type='DenseSampleFrames', clip_len=8, frame_interval=8, num_clips=1),
-        dict(type='FrameSelector', io_backend='disk'),
+        dict(type='RawFrameDecode', io_backend='disk'),
         dict(type='MyTransform'),       # use a custom pipeline
         dict(type='Normalize', **img_norm_cfg),
         dict(type='FormatShape', input_format='NCTHW'),

mmaction/datasets/pipelines/__init__.py

@@ -7,8 +7,8 @@ from .formating import (Collect, FormatShape, ImageToTensor, ToDataContainer,
 from .loading import (DecordDecode, DecordInit, DenseSampleFrames,
                       FrameSelector, GenerateLocalizationLabels,
                       LoadLocalizationFeature, LoadProposals, OpenCVDecode,
-                      OpenCVInit, PyAVDecode, PyAVInit, SampleFrames,
-                      SampleProposalFrames)
+                      OpenCVInit, PyAVDecode, PyAVInit, RawFrameDecode,
+                      SampleFrames, SampleProposalFrames)
 
 __all__ = [
     'SampleFrames', 'PyAVDecode', 'DecordDecode', 'DenseSampleFrames',
@@ -17,5 +17,6 @@ __all__ = [
     'ThreeCrop', 'CenterCrop', 'TenCrop', 'ImageToTensor', 'Transpose',
     'Collect', 'FormatShape', 'Compose', 'ToTensor', 'ToDataContainer',
     'GenerateLocalizationLabels', 'LoadLocalizationFeature', 'LoadProposals',
-    'DecordInit', 'OpenCVInit', 'PyAVInit', 'SampleProposalFrames'
+    'DecordInit', 'OpenCVInit', 'PyAVInit', 'SampleProposalFrames',
+    'RawFrameDecode'
 ]

mmaction/datasets/pipelines/loading.py

@@ -2,6 +2,7 @@ import io
 import os
 import os.path as osp
 import shutil
+import warnings
 
 import mmcv
 import numpy as np
@@ -793,8 +794,8 @@ class OpenCVDecode(object):
 
 
 @PIPELINES.register_module()
-class FrameSelector(object):
-    """Select raw frames with given indices.
+class RawFrameDecode(object):
+    """Load and decode frames with given indices.
 
     Required keys are "frame_dir", "filename_tmpl" and "frame_inds",
     added or modified keys are "imgs", "img_shape" and "original_shape".
@@ -813,7 +814,7 @@ class FrameSelector(object):
         self.file_client = None
 
     def __call__(self, results):
-        """Perform the FrameSelector selecting given indices.
+        """Perform the ``RawFrameDecode`` to pick frames given indices.
 
         Args:
             results (dict): The resulting dict to be modified and passed
@@ -863,6 +864,16 @@ class FrameSelector(object):
         return results
 
 
+@PIPELINES.register_module()
+class FrameSelector(RawFrameDecode):
+    """Deprecated class for ``RawFrameDecode``."""
+
+    def __init__(self, *args, **kwargs):
+        warnings.warn('"FrameSelector" is deprecated, please switch to'
+                      '"RawFrameDecode"')
+        super().__init__(*args, **kwargs)
+
+
 @PIPELINES.register_module()
 class LoadLocalizationFeature(object):
     """Load Video features for localizer with given video_name list.

tests/test_data/test_dataset.py

@@ -37,7 +37,7 @@ class TestDataset(object):
                 clip_len=32,
                 frame_interval=2,
                 num_clips=1),
-            dict(type='FrameSelector', io_backend='disk')
+            dict(type='RawFrameDecode', io_backend='disk')
         ]
         cls.video_pipeline = [
             dict(type='OpenCVInit'),

tests/test_data/test_loading.py

@@ -12,7 +12,8 @@ from mmaction.datasets.pipelines import (DecordDecode, DecordInit,
                                          LoadLocalizationFeature,
                                          LoadProposals, OpenCVDecode,
                                          OpenCVInit, PyAVDecode, PyAVInit,
-                                         SampleFrames, SampleProposalFrames)
+                                         RawFrameDecode, SampleFrames,
+                                         SampleProposalFrames)
 
 
 class ExampleSSNInstance(object):
@@ -830,7 +831,12 @@ class TestLoading(object):
         assert np.shape(opencv_decode_result['imgs']) == (len(
             video_result['frame_inds']), 256, 340, 3)
 
-    def test_frame_selector(self):
+    def test_rawframe_selector(self):
+
+        with pytest.warns(UserWarning):
+            FrameSelector(io_backend='disk')
+
+    def test_rawframe_decode(self):
         target_keys = ['frame_inds', 'imgs', 'original_shape', 'modality']
 
         # test frame selector with 2 dim input when start_index = 0
@@ -840,7 +846,7 @@ class TestLoading(object):
         # since the test images start with index 1, we plus 1 to frame_inds
         # in order to pass the CI
         inputs['frame_inds'] = inputs['frame_inds'] + 1
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -851,7 +857,7 @@ class TestLoading(object):
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(1, self.total_frames, 2)[:,
                                                                   np.newaxis]
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -864,7 +870,7 @@ class TestLoading(object):
         # since the test images start with index 1, we plus 1 to frame_inds
         # in order to pass the CI
         inputs['frame_inds'] = inputs['frame_inds'] + 1
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -874,7 +880,7 @@ class TestLoading(object):
         # test frame selector with 1 dim input
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(1, self.total_frames, 5)
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -887,7 +893,7 @@ class TestLoading(object):
         # since the test images start with index 1, we plus 1 to frame_inds
         # in order to pass the CI
         inputs['frame_inds'] = inputs['frame_inds'] + 1
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -897,7 +903,7 @@ class TestLoading(object):
         # test frame selector with 1 dim input
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(1, self.total_frames, 2)
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']), 240,
@@ -911,7 +917,7 @@ class TestLoading(object):
         # since the test images start with index 1, we plus 1 to frame_inds
         # in order to pass the CI
         inputs['frame_inds'] = inputs['frame_inds'] + 1
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']) * 2,
@@ -921,7 +927,7 @@ class TestLoading(object):
         # test frame selector with 1 dim input for flow images
         inputs = copy.deepcopy(self.flow_frame_results)
         inputs['frame_inds'] = np.arange(1, self.total_frames, 2)
-        frame_selector = FrameSelector(io_backend='disk')
+        frame_selector = RawFrameDecode(io_backend='disk')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
         assert np.shape(results['imgs']) == (len(inputs['frame_inds']) * 2,
@@ -935,7 +941,7 @@ class TestLoading(object):
         # since the test images start with index 1, we plus 1 to frame_inds
         # in order to pass the CI
         inputs['frame_inds'] = inputs['frame_inds'] + 1
-        frame_selector = FrameSelector(
+        frame_selector = RawFrameDecode(
             io_backend='disk', decoding_backend='turbojpeg')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)
@@ -946,7 +952,7 @@ class TestLoading(object):
         # test frame selector in turbojpeg decording backend
         inputs = copy.deepcopy(self.frame_results)
         inputs['frame_inds'] = np.arange(1, self.total_frames, 5)
-        frame_selector = FrameSelector(
+        frame_selector = RawFrameDecode(
             io_backend='disk', decoding_backend='turbojpeg')
         results = frame_selector(inputs)
         assert self.check_keys_contain(results.keys(), target_keys)