Skeleton-based Action Recognition on NTU RGB+D 60 (Cross-Subject)
97AccuracySkeleT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SkeleTGFLOPs=9.6, Params(M)=5.22026.01 | 97 | — | |
| Ours (BHaRNet-P)variant=probabilistic, GFLOPs=6.8, Params(M)=9.72026.01 | 96.8 | — | |
| Ours (BHaRNet-E)variant=probabilistic, GFLOPs=10.9, Params(M)=11.02026.01 | 96.7 | — | |
| Ours (BHaRNet-B)variant=probabilistic, GFLOPs=6.6, Params(M)=5.52026.01 | 96.6 | — | |
| BHaRNet-P†GFLOPs=3.4, Params(M)=4.92026.01 | 96.3 | — | |
| BHaRNet-E†GFLOPs=5.4, Params(M)=5.52026.01 | 96.2 | — | |
| BHaRNet-B†GFLOPs=3.3, Params(M)=2.82026.01 | 96.1 | — | |
| 3MformerGFLOPs=58.5, Params(M)=6.72026.01 | 94.8 | — | |
| PoseConv3DGFLOPs=31.8, Params(M)=4.02026.01 | 94.1 | — | |
| ProtoGCNGFLOPs=43.4, Params(M)=24.92026.01 | 93.8 | — | |
| DeGCNGFLOPs=6.9, Params(M)=5.62026.01 | 93.6 | — | |
| BlockGCNGFLOPs=6.5, Params(M)=5.22026.01 | 93.1 | — | |
| InfoGCNGFLOPs=10.0*, Params(M)=9.42026.01 | 93 | — | |
| CTR-GCNGFLOPs=7.9, Params(M)=5.82026.01 | 92.4 | — | |
| AGE-EnsInput=S(J)&S(B)&S(J+B)&V(J+B), # Ens=4, # Params (M)=5.80, GFlops=78.02021.05 | 91.6 | — | |
| MS-G3DYear=2020, # Ens=2, # Params (M)=6.44, GFlops=98.02021.05 | 91.5 | — | |
| AGE-EnsInput=S(B)&S(J+B)&V(J+B), # Ens=3, # Params (M)=4.36, GFlops=58.62021.05 | 91.4 | — | |
| MSTYear=2021, # Ens=22021.05 | 91.1 | — | |
| AGE-EnsInput=S(J+B)&V(J+B), # Ens=2, # Params (M)=2.92, GFlops=39.22021.05 | 91 | — | |
| BSL-EnsInput=S(J)&S(B)&S(J+B)&V(J+B), # Ens=4, # Params (M)=5.72, GFlops=76.82021.05 | 90.9 | — | |
| DeCoupleGCNYear=2020, # Ens=4, # Params (M)=13.72, GFlops=102.32021.05 | 90.8 | — | |
| AGE-EnsInput=S(B)&V(B), # Ens=2, # Params (M)=2.88, GFlops=38.82021.05 | 90.8 | — | |
| Ta-CNNYear=2022, # Ens=22021.05 | 90.7 | — | |
| BSL-EnsInput=S(B)&S(J+B)&V(J+B), # Ens=3, # Params (M)=4.30, GFlops=57.82021.05 | 90.7 | — | |
| AdaSGNYear=2021, # Ens=42021.05 | 90.5 | — | |
| Efficient-Self-AttentionYear=2022, # Ens=42021.05 | 90.5 | — | |
| AGE-EnsInput=S(J)&V(J), # Ens=2, # Params (M)=2.88, GFlops=38.82021.05 | 90.5 | — | |
| BSL-EnsInput=S(B)&V(B), # Ens=2, # Params (M)=2.84, GFlops=38.02021.05 | 90.5 | — | |
| BSL-EnsInput=S(J+B)&V(J+B), # Ens=2, # Params (M)=2.88, GFlops=38.82021.05 | 90.5 | — | |
| AGE-SInput=Joint+Bone, # Ens=1, # Params (M)=1.46, GFlops=19.62021.05 | 90 | — | |
| DGNNYear=2019, # Ens=4, # Params (M)=8.06, GFlops=71.12021.05 | 89.9 | — | |
| BSL-EnsInput=S(J)&V(J), # Ens=2, # Params (M)=2.84, GFlops=38.02021.05 | 89.3 | — | |
| AGC-LSTMYear=2019, # Ens=22021.05 | 89.2 | — | |
| AGE-SInput=Bone, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 89.2 | — | |
| BSL-SInput=Joint+Bone, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 89.2 | — | |
| SGNYear=2020, # Ens=1, # Params (M)=0.69, GFlops=15.42021.05 | 89 | — | |
| AGE-SInput=Joint, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 88.7 | — | |
| 2s-AGCNYear=2019, # Ens=4, # Params (M)=6.72, GFlops=37.22021.05 | 88.5 | — | |
| BSL-SInput=Bone, # Ens=1, # Params (M)=1.42, GFlops=19.02021.05 | 88.2 | — | |
| AGE-VInput=Joint, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 88.2 | — | |
| AGE-VInput=Bone, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 88 | — | |
| BSL-SInput=Joint, # Ens=1, # Params (M)=1.42, GFlops=19.02021.05 | 87.2 | — | |
| AGE-VInput=Joint+Bone, # Ens=1, # Params (M)=1.46, GFlops=19.62021.05 | 87.1 | — | |
| AS-GCNYear=2019, # Ens=1, # Params (M)=7.17, GFlops=35.52021.05 | 86.8 | — | |
| HCNYear=2018, # Ens=12021.05 | 86.5 | — | |
| BSL-VInput=Bone, # Ens=1, # Params (M)=1.42, GFlops=19.02021.05 | 86.4 | — | |
| BSL-VInput=Joint+Bone, # Ens=1, # Params (M)=1.44, GFlops=19.42021.05 | 86.1 | — | |
| BSL-VInput=Joint, # Ens=1, # Params (M)=1.42, GFlops=19.02021.05 | 86 | — | |
| MANYear=2018, # Ens=12021.05 | 82.7 | — | |
| Bayes-GCNYear=2019, # Ens=12021.05 | 81.8 | — | |
| ST-GCNYear=2018, # Ens=1, # Params (M)=2.91, GFlops=16.42021.05 | 81.5 | — | |
| 3s-CrosSCLREvaluation Protocol=Linear2026.01 | 77.8 | — | |
| GL-TransformerEvaluation Protocol=Linear2026.01 | 76.3 | — | |
| Variational Contrastive Learning (VCL)Evaluation Protocol=Linear, Backbone=ST-GCN2026.01 | 75.2 | — | |
| TS-ColorizationEvaluation Protocol=Linear2026.01 | 71.6 | — | |
| AS-CALEvaluation Protocol=Linear2026.01 | 58.5 | — | |
| MS2LEvaluation Protocol=Linear2026.01 | 52.6 | — | |
| P&CEvaluation Protocol=Linear2026.01 | 50.7 | — | |
| LongT GANEvaluation Protocol=Linear2026.01 | 39.1 | — | |
| 2s-AGCNPublication=CVPR20192023.03 | — | 88.5 | |
| 2s-AGCN2021.08 | — | 88.9 | |
| AGC-LSTMPublication=CVPR20192023.03 | — | 89.2 | |
| AS-GCN2021.08 | — | 86.8 | |
| CTR-GCNPublication=ICCV20212023.03 | — | 92.4 | |
| DC-GCN+ADGPublication=ECCV20202023.03 | — | 90.8 | |
| DDGCNPublication=ECCV20202023.03 | — | 91.1 | |
| DecoupleGCN2021.08 | — | 90.8 | |
| DGNNPublication=CVPR20192023.03 | — | 89.9 | |
| Dynamic GCNPublication=ACMMM20202023.03 | — | 91.5 | |
| EfficientGCN-B4Publication=TPAMI20222023.03 | — | 91.7 | |
| FR HeadFusion=joint, bone, joint motion, and bone motion2023.03 | — | 92.8 | |
| HCSFframework=multi-stream2021.08 | — | 91.6 | |
| Ind-RNNPublication=CVPR20182023.03 | — | 81.8 | |
| Mix-dim2021.08 | — | 89.7 | |
| MMDGCN2021.08 | — | 90.8 | |
| MS-AAGCN2021.08 | — | 90 | |
| MS-G3DPublication=CVPR20202023.03 | — | 91.5 | |
| MS-G3D2021.08 | — | 91.5 | |
| MST-GCNPublication=AAAI20212023.03 | — | 91.5 | |
| NAS-GCN2021.08 | — | 89.4 | |
| PA-ResGCN-B19Publication=ACMMM20202023.03 | — | 90.9 | |
| SGNPublication=CVPR20202023.03 | — | 89 | |
| SGN2021.08 | — | 89 | |
| Shift-GCNPublication=CVPR20202023.03 | — | 90.7 | |
| Shift-GCN2021.08 | — | 90.7 | |
| Skeletal-GNNPublication=ICCV20212023.03 | — | 91.6 | |
| ST-GCNPublication=AAAI20182023.03 | — | 81.5 | |
| ST-GCN2021.08 | — | 84.3 | |
| ST-Transformer2021.08 | — | 89.3 | |
| STFPublication=AAAI20222023.03 | — | 92.5 | |
| Ta-CNNPublication=AAAI20222023.03 | — | 90.4 |