Skeleton-based Action Recognition on NTU (Cross-Subject)
89.4AccuracyOurs(Joint+Bone)
Evaluation Results
| Method | Links | |
|---|---|---|
| Ours(Joint+Bone)2019.11 | 89.4 | |
| 2S-AGCNConference=CVPR20192019.11 | 88.5 | |
| Complete GR-GCN modelGraph Construction=Physical + Non-physical + Temporal2018.11 | 87.5 | |
| GR-GCNConference=ACMM20192019.11 | 87.5 | |
| STGR-GCNYear=20192018.11 | 86.9 | |
| STGR-GCNConference=AAAI20192019.11 | 86.9 | |
| AS-GCNConference=CVPR20192019.11 | 86.8 | |
| GR-GCN (Bone + Intra-connection)Graph Construction=Physical + Non-physical (non-temporal)2018.11 | 85 | |
| SR-TSL2018.05 | 84.8 | |
| SR-TSLYear=20182018.11 | 84.8 | |
| SR-TSLConference=ECCV20182019.11 | 84.8 | |
| GR-GCN (Bone only)Graph Construction=Physical connections only (Baseline)2018.11 | 84.3 | |
| DPRLYear=20182018.11 | 83.5 | |
| DPRLConference=CVPR20182019.11 | 83.5 | |
| Two-Stream CNNYear=20172018.11 | 83.2 | |
| LSTM-CNNYear=20172018.11 | 82.9 | |
| IndRNNLayers=62018.03 | 81.8 | |
| ST-GCN2018.05 | 81.5 | |
| ST-GCNYear=20182018.11 | 81.5 | |
| ST-GCNConference=AAAI20182019.11 | 81.5 | |
| EleAtt-GRUData Augmentation=rotating skeleton sequences2018.07 | 80.7 | |
| EleAtt-GRUdata augmentation=true2019.09 | 80.7 | |
| Enhanced Visualization+CNN2018.03 | 80.03 | |
| SynCNNConference=PR2017(Jou.)2019.11 | 80 | |
| EleAtt-GRU2018.07 | 79.8 | |
| EleAtt-GRUdata augmentation=false2019.09 | 79.8 | |
| Clips + CNN + MTLN2018.07 | 79.6 | |
| Clips+CNN+MTLN2019.09 | 79.6 | |
| Clips+CNN+MTLN2018.03 | 79.57 | |
| VA-LSTM2018.07 | 79.4 | |
| VA-LSTM2018.05 | 79.4 | |
| VA-LSTM2019.09 | 79.4 | |
| VA-LSTMConference=CVPR20172019.11 | 79.2 | |
| IndRNNLayers=42018.03 | 78.58 | |
| Pose conditioned STA-LSTM2018.03 | 77.1 | |
| JDM+CNN2018.03 | 76.2 | |
| Visualization CNN2018.05 | 76 | |
| SkeletonNet(CNN)2018.03 | 75.94 | |
| Baseline-GRULayers=32018.07 | 75.2 | |
| Baseline-GRU2019.09 | 75.2 | |
| Deep STGCKYear=20182018.11 | 74.9 | |
| Deep STGCKConference=AAAI20182019.11 | 74.9 | |
| URNN-2L-T2018.07 | 74.6 | |
| Ensemble TS-LSTM2018.05 | 74.6 | |
| URNN-2L-T2019.09 | 74.6 | |
| GCA-LSTM2018.07 | 74.4 | |
| GCA-LSTM2019.09 | 74.4 | |
| Res-TCN2018.03 | 74.3 | |
| TCNConference=CVPRW20172019.11 | 74.3 | |
| STA-LSTM2018.07 | 73.4 | |
| STA-LSTM2018.05 | 73.4 | |
| JTM+CNN2018.03 | 73.4 | |
| STA-LSTM2018.03 | 73.4 | |
| STA-LSTMConference=AAAI20172019.11 | 73.4 | |
| STA-LSTM2019.09 | 73.4 | |
| Two-stream RNN2018.05 | 71.3 | |
| Geometric FeaturesYear=20172018.11 | 70.3 | |
| JL_d+RNN2018.03 | 70.26 | |
| ST-LSTM + Trust Gate2018.07 | 69.2 | |
| Trust Gate ST-LSTM2018.05 | 69.2 | |
| ST-LSTM (Tree)+Trust GateYear=20182018.11 | 69.2 | |
| ST-LSTM + Trust Gate2018.03 | 69.2 | |
| ST-LSTM + Trust Gate2019.09 | 69.2 | |
| PLSTMLayers=22018.03 | 62.93 | |
| Part-aware LSTM2018.07 | 62.9 | |
| Part-aware LSTM2018.05 | 62.9 | |
| Part-aware LSTMYear=20162018.11 | 62.9 | |
| P-LSTMConference=CVPR20162019.11 | 62.9 | |
| Part-aware LSTM2019.09 | 62.9 | |
| PLSTMLayers=12018.03 | 62.05 | |
| Deep learning on Lie Group2018.03 | 61.37 | |
| Dynamic Skeletons2018.07 | 60.2 | |
| Dynamic SkeletonsYear=20152018.11 | 60.2 | |
| JointConference=CVPR20152019.11 | 60.2 | |
| Dynamic Skeletons2019.09 | 60.2 | |
| LSTMLayers=22018.03 | 60.09 | |
| LSTMLayers=12018.03 | 59.14 | |
| HBRNN-L2018.07 | 59.1 | |
| HBRNN-L2018.05 | 59.1 | |
| HBRNN-L2019.09 | 59.1 | |
| RNNLayers=22018.03 | 56.29 | |
| RNNLayers=12018.03 | 56.02 | |
| Lie Group2018.07 | 50.1 | |
| Lie Group2019.09 | 50.1 | |
| Skeleton Quads2018.07 | 38.6 | |
| Skeleton Quads2019.09 | 38.6 |