Upload 41 files

ntu60_xview/b_2/b_2.py

@@ -0,0 +1,93 @@
+modality = 'b'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/b_2'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_1',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_3', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='GenSkeFeat', feats=['b']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['b']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['b']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='GenSkeFeat', feats=['b']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['b']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['b']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/b_2/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e034cfe4c881439b1c7f23e26255cdcd93b851e62fba5e2d58b994895df0d35d
+size 7622053

ntu60_xview/b_2/best_top1_acc_epoch_148.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0b559a3b17ddad1908dec8ffd1c769c8d46b58f1897317bc91f7098dd2778f2
+size 33472486

ntu60_xview/bm/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a34369d7eda12da401c4b71763f7a169e87aebc94001b131ae2cc40b42b7112
+size 7623135

ntu60_xview/bm/best_top1_acc_epoch_139.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dba64661aebf7b198f84f589fbb250e405e155add8cfd7595fd0714abf181bc
+size 33472486

ntu60_xview/bm/bm.py

@@ -0,0 +1,93 @@
+modality = 'bm'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/bm'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_1',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_3', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='GenSkeFeat', feats=['bm']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['bm']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['bm']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='GenSkeFeat', feats=['bm']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['bm']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['bm']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/j_1/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0774e5049a5b3cffb1b7d174a7902060457cc7ca4125618a3d8bc447d0543f7e
+size 7622891

ntu60_xview/j_1/best_top1_acc_epoch_145.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:576e5287e7b630b385e7539566dc0cee1d6093282d21e5d12a59482b01f88b1b
+size 35315686

ntu60_xview/j_1/j_1.py

@@ -0,0 +1,93 @@
+modality = 'j'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/j_1'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_2',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_2', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/j_2/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff103f0cd3bd2454b1dc33ae7dd6867cf0e6b1feb6bb9b2783423b69b0f3f038
+size 7622150

ntu60_xview/j_2/best_top1_acc_epoch_141.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da876749f60ac3367f94417d555b993b9382814c0aeab35962552eb72181fa74
+size 35315686

ntu60_xview/j_2/j_2.py

@@ -0,0 +1,93 @@
+modality = 'j'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/j_2'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_2',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_2', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['j']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['j']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/jm/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66f3bb6c194c3b00688da5e4a942281ae9900ea9467d2b69996eca2cd37661ae
+size 7624332

ntu60_xview/jm/best_top1_acc_epoch_149.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19ad1b76408427507ba0460040b8a15d06abba79603cff2bcbfca9eb1e90f1f3
+size 35315686

ntu60_xview/jm/jm.py

@@ -0,0 +1,93 @@
+modality = 'jm'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/jm'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_2',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_2', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='GenSkeFeat', feats=['jm']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['jm']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['jm']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='GenSkeFeat', feats=['jm']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['jm']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['jm']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/k_1/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39857ed408fb9d8bc01f57e1a2b1cb11f15c977e36b619e532991b423b237f92
+size 7621133

ntu60_xview/k_1/best_top1_acc_epoch_145.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e626a8b75d81b97690b5dbf8899b64c15977fa1d9d1f894dae1bb914c39f03a
+size 33472486

ntu60_xview/k_1/k_1.py

@@ -0,0 +1,95 @@
+modality = 'k'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/k_1'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_1',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_1', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='Part_Drop'),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='Part_Drop'),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/k_2/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d645501b9dac84f9da25574b54a2f0fdfafabbe6c93451db11e294b3667dfab0
+size 7622841

ntu60_xview/k_2/best_top1_acc_epoch_147.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5cff77b348e85f8688c7ab6d085921c792f6fffb9b6722724fd807b7b159709f
+size 33472486

ntu60_xview/k_2/k_2.py

@@ -0,0 +1,95 @@
+modality = 'k'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/k_2'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_1',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_1', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='Part_Drop'),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['k']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='Part_Drop'),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['k']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/km/best_pred.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40dc4cdc42e5451f4770fea8653ef782bfbd8663f0883086553b4ad7d0adce10
+size 7621999

ntu60_xview/km/best_top1_acc_epoch_145.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a0b04085f59bbcc1f350916f17d96f6bae9d9aa6b1d78dd5ab3973c346528cf
+size 33472486

ntu60_xview/km/km.py

@@ -0,0 +1,95 @@
+modality = 'km'
+graph = 'nturgb+d'
+work_dir = './work_dirs/test_prototype/ntu60_xview/km'
+model = dict(
+    type='RecognizerGCN_7_1_1',
+    backbone=dict(
+        type='GCN_7_1_1',
+        tcn_ms_cfg=[(3, 1), (3, 2), (3, 3), (3, 4), ('max', 3), '1x1'],
+        graph_cfg=dict(
+            layout='nturgb+d',
+            mode='random',
+            num_filter=8,
+            init_off=0.04,
+            init_std=0.02)),
+    cls_head=dict(type='SimpleHead_7_4_1', num_classes=60, in_channels=384))
+dataset_type = 'PoseDataset'
+ann_file = '/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl'
+train_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='RandomRot', theta=0.2),
+    dict(type='Part_Drop'),
+    dict(type='GenSkeFeat', feats=['km']),
+    dict(type='UniformSampleDecode', clip_len=100),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+val_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['km']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+test_pipeline = [
+    dict(type='PreNormalize3D', align_spine=False),
+    dict(type='GenSkeFeat', feats=['km']),
+    dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+    dict(type='FormatGCNInput'),
+    dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+    dict(type='ToTensor', keys=['keypoint'])
+]
+data = dict(
+    videos_per_gpu=16,
+    workers_per_gpu=4,
+    test_dataloader=dict(videos_per_gpu=1),
+    train=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='RandomRot', theta=0.2),
+            dict(type='Part_Drop'),
+            dict(type='GenSkeFeat', feats=['km']),
+            dict(type='UniformSampleDecode', clip_len=100),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_train'),
+    val=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['km']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=1),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'),
+    test=dict(
+        type='PoseDataset',
+        ann_file='/data/lhd/pyskl_data/nturgbd/ntu60_3danno.pkl',
+        pipeline=[
+            dict(type='PreNormalize3D', align_spine=False),
+            dict(type='GenSkeFeat', feats=['km']),
+            dict(type='UniformSampleDecode', clip_len=100, num_clips=10),
+            dict(type='FormatGCNInput'),
+            dict(type='Collect', keys=['keypoint', 'label'], meta_keys=[]),
+            dict(type='ToTensor', keys=['keypoint'])
+        ],
+        split='xview_val'))
+optimizer = dict(
+    type='SGD', lr=0.025, momentum=0.9, weight_decay=0.0005, nesterov=True)
+optimizer_config = dict(grad_clip=None)
+lr_config = dict(policy='CosineAnnealing', min_lr=0, by_epoch=False)
+total_epochs = 150
+checkpoint_config = dict(interval=1)
+evaluation = dict(interval=1, metrics=['top_k_accuracy'])
+log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
+dist_params = dict(backend='nccl')
+gpu_ids = range(0, 1)

ntu60_xview/ntu60_xview_ensemble.py

@@ -0,0 +1,65 @@
+from mmcv import load
+import sys
+# Note: please adjust the relative path according to the actual situation.
+sys.path.append('../..')
+from protogcn.smp import *
+
+
+j_1 = load('j_1/best_pred.pkl')
+b_1 = load('b_1/best_pred.pkl')
+k_1 = load('k_1/best_pred.pkl')
+j_2 = load('j_2/best_pred.pkl')
+b_2 = load('b_2/best_pred.pkl')
+k_2 = load('k_2/best_pred.pkl')
+jm = load('jm/best_pred.pkl')
+bm = load('bm/best_pred.pkl')
+km = load('km/best_pred.pkl')
+label = load_label('/data/nturgbd/ntu60_3danno.pkl', 'xview_val')
+
+
+"""
+***************
+InfoGCN  v0:
+j jm b bm k km
+2S:     97.23
+4S:     97.42
+6S:     97.67
+***************
+"""
+print('InfoGCN  v0:')
+print('j jm b bm k km')
+print('2S')
+fused = comb([j_1, b_1], [1, 1])
+print('Top-1', top1(fused, label))
+
+print('4S')
+fused = comb([j_1, b_1, jm, bm], [2, 2, 1, 1])
+print('Top-1', top1(fused, label))
+
+print('6S')
+fused = comb([j_1, b_1, k_1, jm, bm, km], [5, 5, 5, 3, 3, 3])
+print('Top-1', top1(fused, label))
+
+
+"""
+***************
+InfoGCN  v1:
+j j b b k k
+2S:     97.23
+4S:     97.49
+6S:     97.76
+***************
+"""
+print('InfoGCN  v1:')
+print('j j b b k k')
+print('2S')
+fused = comb([j_1, b_1], [1, 1])
+print('Top-1', top1(fused, label))
+
+print('4S')
+fused = comb([j_1, b_1, j_2, b_2], [1, 1, 2, 2])
+print('Top-1', top1(fused, label))
+
+print('6S')
+fused = comb([j_1, j_2, b_1, b_2, k_1, k_2], [5, 5, 6, 6, 6, 6])
+print('Top-1', top1(fused, label))