# Action Recognition on NTU120 (cross-subject (CS)) Source: [Action Recognition on NTU120 (cross-subject (CS))]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:07:07.215Z Evaluation of video action recognition performance on the NTU-120 dataset using the standard Cross-Subject (CS) evaluation protocol, measured by Top-1 accuracy. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Action Recognition | | Dataset (including any stated split) | NTU120 (cross-subject (CS)) | | Metric | Top-1 Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 36 of 36 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Top-1 Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Top-1 Accuracy | Result date | Report paper | | --- | --- | --- | --- | | PoseC3D | 95.3 | Not recorded | 68fa715c1e209131cea50dad | | π-ViT + 3D Poses | 95.1 | Not recorded | 68fa715c1e209131cea50dad | | 3Mformer | 92 | Not recorded | 68fa715c1e209131cea50dad | | π-ViT | 91.9 | Not recorded | 68fa715c1e209131cea50dad | | + 3D-SIM | 91.8 | Not recorded | 68fa715c1e209131cea50dad | | Video Swin | 91.4 | Not recorded | 68fa715c1e209131cea50dad | | VPN++ + 3D Poses | 90.7 | Not recorded | 68fa715c1e209131cea50dad | | TimeSformer | 90.6 | Not recorded | 68fa715c1e209131cea50dad | | 3D-Def-Transformer | 90.5 | Not recorded | 68fa715c1e209131cea50dad | | + 2D-SIM | 90.5 | Not recorded | 68fa715c1e209131cea50dad | | DG-STGCN(A0.66) | 90.4 | Not recorded | 68fa789d1e209131cea795af | | STAR-Transformer | 90.3 | Not recorded | 68fa715c1e209131cea50dad | | InfoGCN | 89.8 | Not recorded | 68fa715c1e209131cea50dad | | DG-STGCN | 89.6 | Not recorded | 68fa789d1e209131cea795af | | DG-STGCN(S0.6) | 89.1 | Not recorded | 68fa789d1e209131cea795af | | DG-STGCN(S0.8) | 89 | Not recorded | 68fa789d1e209131cea795af | | CTR-GCN | 88.9 | Not recorded | 68fa789d1e209131cea795af | | DG-STGCN(S0.95) | 88.7 | Not recorded | 68fa789d1e209131cea795af | | ST-GCN++ | 88.6 | Not recorded | 68fa789d1e209131cea795af | | SKE2GRID | 87.3 | Not recorded | 68fa789d1e209131cea795af | | MotionFormer | 87 | Not recorded | 68fa715c1e209131cea50dad | | MS-G3D | 86.9 | Not recorded | 68fa789d1e209131cea795af | | PoseConv3D | 86.9 | Not recorded | 68fa789d1e209131cea795af | | VPN++ | 86.7 | Not recorded | 68fa715c1e209131cea50dad | | Hyperformer | 86.6 | Not recorded | 68fa715c1e209131cea50dad | | DSTA-Net | 86.6 | Not recorded | 68fa789d1e209131cea795af | | VPN | 86.3 | Not recorded | 68fa715c1e209131cea50dad | | ViewCLR | 86.2 | Not recorded | 68fa715c1e209131cea50dad | | PoseC3D | 86 | Not recorded | 68fa715c1e209131cea50dad | | ShiftGCN | 85.9 | Not recorded | 68fa789d1e209131cea795af | | ViewCon | 85.6 | Not recorded | 68fa715c1e209131cea50dad | | FGCN | 85.4 | Not recorded | 68fa789d1e209131cea795af | | DG-STGCN(S0.99) | 85.3 | Not recorded | 68fa789d1e209131cea795af | | AS-GCN | 78.3 | Not recorded | 68fa789d1e209131cea795af | | RA-GCN | 78.3 | Not recorded | 68fa789d1e209131cea795af | | ST-GCN | 70.7 | Not recorded | 68fa789d1e209131cea795af | ### PoseC3D Top-1 Accuracy: 95.3. Result ID: 6991c9813a6c3148490a9426. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### π-ViT + 3D Poses Top-1 Accuracy: 95.1. Result ID: 6991c9813a6c3148490a9429. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### 3Mformer Top-1 Accuracy: 92. Result ID: 6991c9813a6c3148490a9423. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Not used"}. This record does not assert independent reproduction. ### π-ViT Top-1 Accuracy: 91.9. Result ID: 6991c9813a6c3148490a9432. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used only in training","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### + 3D-SIM Top-1 Accuracy: 91.8. Result ID: 6991c9813a6c3148490a9431. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used only in training","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### Video Swin Top-1 Accuracy: 91.4. Result ID: 6991c9813a6c3148490a942d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Not used","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### VPN++ + 3D Poses Top-1 Accuracy: 90.7. Result ID: 6991c9813a6c3148490a9425. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### TimeSformer Top-1 Accuracy: 90.6. Result ID: 6991c9813a6c3148490a942f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Not used","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### 3D-Def-Transformer Top-1 Accuracy: 90.5. Result ID: 6991c9813a6c3148490a9428. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### + 2D-SIM Top-1 Accuracy: 90.5. Result ID: 6991c9813a6c3148490a9430. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used only in training","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### DG-STGCN(A0.66) Top-1 Accuracy: 90.4. Result ID: 6995959b3badfaffe6905b78. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DG-STGCN","Sparsity level":"0.66","Mode":"Multi-level Assembly"}. This record does not assert independent reproduction. ### STAR-Transformer Top-1 Accuracy: 90.3. Result ID: 6991c9813a6c3148490a9427. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### InfoGCN Top-1 Accuracy: 89.8. Result ID: 6991c9813a6c3148490a9420. [Reporting paper]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Not used"}. This record does not assert independent reproduction. ### DG-STGCN Top-1 Accuracy: 89.6. Result ID: 6995959b3badfaffe6905b73. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DG-STGCN"}. This record does not assert independent reproduction. ### DG-STGCN(S0.6) Top-1 Accuracy: 89.1. Result ID: 6995959b3badfaffe6905b74. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DG-STGCN","Sparsity level":"0.6","Mode":"Sparse"}. This record does not assert independent reproduction. ### DG-STGCN(S0.8) Top-1 Accuracy: 89. Result ID: 6995959b3badfaffe6905b75. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DG-STGCN","Sparsity level":"0.8","Mode":"Sparse"}. This record does not assert independent reproduction. ### CTR-GCN Top-1 Accuracy: 88.9. Result ID: 6995959b3badfaffe6905b6f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"CTR-GCN"}. This record does not assert independent reproduction. ### DG-STGCN(S0.95) Top-1 Accuracy: 88.7. Result ID: 6995959b3badfaffe6905b76. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DG-STGCN","Sparsity level":"0.95","Mode":"Sparse"}. This record does not assert independent reproduction. ### ST-GCN++ Top-1 Accuracy: 88.6. Result ID: 6995959b3badfaffe6905b70. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ST-GCN++"}. This record does not assert independent reproduction. ### SKE2GRID Top-1 Accuracy: 87.3. Result ID: 6995959b3badfaffe6905b72. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"SKE2GRID"}. This record does not assert independent reproduction. ### MotionFormer Top-1 Accuracy: 87. Result ID: 6991c9813a6c3148490a942e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Not used","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### MS-G3D Top-1 Accuracy: 86.9. Result ID: 6995959b3badfaffe6905b6e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"MS-G3D"}. This record does not assert independent reproduction. ### PoseConv3D Top-1 Accuracy: 86.9. Result ID: 6995959b3badfaffe6905b71. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"PoseConv3D"}. This record does not assert independent reproduction. ### VPN++ Top-1 Accuracy: 86.7. Result ID: 6991c9813a6c3148490a942a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used only in training","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### Hyperformer Top-1 Accuracy: 86.6. Result ID: 6991c9813a6c3148490a9422. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Not used"}. This record does not assert independent reproduction. ### DSTA-Net Top-1 Accuracy: 86.6. Result ID: 6995959b3badfaffe6905b6d. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"DSTA-Net"}. This record does not assert independent reproduction. ### VPN Top-1 Accuracy: 86.3. Result ID: 6991c9813a6c3148490a9424. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### ViewCLR Top-1 Accuracy: 86.2. Result ID: 6991c9813a6c3148490a942b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Modality (Pose)":"Not used","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### PoseC3D Top-1 Accuracy: 86. Result ID: 6991c9813a6c3148490a9421. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Modality (Pose)":"Used at inference","Modality (RGB)":"Not used"}. This record does not assert independent reproduction. ### ShiftGCN Top-1 Accuracy: 85.9. Result ID: 6995959b3badfaffe6905b6c. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"ShiftGCN"}. This record does not assert independent reproduction. ### ViewCon Top-1 Accuracy: 85.6. Result ID: 6991c9813a6c3148490a942c. [Reporting paper]() Recorded evaluation configuration: {"Modality (Pose)":"Not used","Modality (RGB)":"Used at inference"}. This record does not assert independent reproduction. ### FGCN Top-1 Accuracy: 85.4. Result ID: 6995959b3badfaffe6905b6b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"FGCN"}. This record does not assert independent reproduction. ### DG-STGCN(S0.99) Top-1 Accuracy: 85.3. Result ID: 6995959b3badfaffe6905b77. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DG-STGCN","Sparsity level":"0.99","Mode":"Sparse"}. This record does not assert independent reproduction. ### AS-GCN Top-1 Accuracy: 78.3. Result ID: 6995959b3badfaffe6905b69. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"AS-GCN"}. This record does not assert independent reproduction. ### RA-GCN Top-1 Accuracy: 78.3. Result ID: 6995959b3badfaffe6905b6a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"RA-GCN"}. This record does not assert independent reproduction. ### ST-GCN Top-1 Accuracy: 70.7. Result ID: 6995959b3badfaffe6905b68. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ST-GCN"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.