Action Recognition on NTU RGB+D 120 (110/10)
80.7AccuracyFlora
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FloraBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 80.7 | 59.8 | 70.5 | 64.7 | |
| FloraBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 79.6 | 66.2 | 66 | 66.1 | |
| DVTABackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 74.9 | — | — | — | |
| InfoCPLBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 74.8 | — | — | — | |
| ScoPLeBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 74.5 | 63.5 | 61.1 | 62.3 | |
| FS-VAEBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 74.4 | 59.2 | 67.9 | 63.3 | |
| TDSMBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 74.2 | — | — | — | |
| PURLSBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 72 | — | — | — | |
| STAR++Backbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 72 | 59 | 55.4 | 57.2 | |
| NeuronBackbone Features=STAR-based, Fusion Type=Two-stream2025.11 | 71.5 | 67.6 | 59.5 | 63.3 | |
| GZSSARBackbone Features=SynSE-based, Fusion Type=Single-stream2025.11 | 71.2 | 46.8 | 68.3 | 55.6 | |
| SA-DAVEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 68.8 | 61.1 | 59.8 | 60.4 | |
| SA-DVAESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 68.77 | — | — | — | |
| SMIESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 65.74 | — | — | — | |
| STARBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 63.3 | 59.9 | 52.7 | 56.1 | |
| SynSESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 62.69 | — | — | — | |
| SMIEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 61.3 | — | — | — | |
| CADA-VAESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 59.53 | — | — | — | |
| JPoSEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 57.3 | 53.6 | 11.6 | 19.1 | |
| ReViSESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 55.04 | — | — | — | |
| CADA-VAEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 52.5 | 50.2 | 43.9 | 46.8 | |
| SynSEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 52.4 | 57.3 | 43.2 | 49.5 | |
| JPOSESkeleton Feature Extractor=Shift-GCN [4], Text Feature Extractor=Sentence-BERT [21], Learning rate=3.48e-05, Batch size=32, Optimizer=Adam2024.07 | 51.93 | — | — | — | |
| ReViSEBackbone Features=STAR-based, Fusion Type=Single-stream2025.11 | 19.8 | 0.6 | 14.5 | 1.1 |