Action Recognition on UCF101 (Top-1 Accuracy)
81.1Top-1 AccuracyTP-CLIP
Evaluation Results
| Method | Links | |
|---|---|---|
| TP-CLIPBackbone=CLIP, Zero-shot=true2026.03 | 81.1 | |
| SP-CLIPBackbone=CLIP, Zero-shot=true2026.03 | 80.4 | |
| EZ-CLIPBackbone=CLIP, Zero-shot=true2026.03 | 79.4 | |
| SDRBackbone=I3D, Zero-shot=true2026.03 | 62.9 | |
| GILBackbone=I3D, Zero-shot=true2026.03 | 55.8 | |
| CLASTERBackbone=I3D, Zero-shot=true2026.03 | 52.7 | |
| ER-enhanced ZSAR modelVideo=(S+Obj)+, Class=ED2021.08 | 51.8 | |
| ER-ZSARBackbone=TSM, Zero-shot=true2026.03 | 51.8 | |
| E2EVideo=r(2+1)d*, Class=WN2021.08 | 48 | |
| E2EBackbone=C3D, Zero-shot=true2026.03 | 48 | |
| OD + SPOTBackbone=I3D, Zero-shot=true2026.03 | 40.9 | |
| RGSCLBackbone=I3D, Zero-shot=true2026.03 | 40.2 | |
| SupVFDBackbone=I3D, Zero-shot=true2026.03 | 39.6 | |
| TS-GCNVideo=Obj, Class=WN2021.08 | 34.2 | |
| O2AVideo=Obj, Class=WN2021.08 | 30.3 | |
| ODBackbone=I3D, Zero-shot=true2026.03 | 26.9 | |
| ASRVideo=C3D*, Class=WT2021.08 | 24.4 | |
| GGMBackbone=C3D, Zero-shot=true2026.03 | 20.3 | |
| URVideo=FV, Class=WN2021.08 | 17.5 | |
| IAPVideo=FV, Class=A2021.08 | 16.7 | |
| DAPVideo=FV, Class=A2021.08 | 15.9 | |
| MTEVideo=FV, Class=WN2021.08 | 15.8 | |
| ZSECOCVideo=FV, Class=WN2021.08 | 15.1 | |
| ESZSLVideo=FV, Class=WN2021.08 | 15 | |
| HAAVideo=FV, Class=A2021.08 | 14.9 | |
| SJEVideo=FV, Class=A2021.08 | 12 | |
| SVEVideo=BoW, Class=WN2021.08 | 10.9 | |
| SJEVideo=FV, Class=WN2021.08 | 9.9 |