Video Recognition on UCF101
98.3AccuracyViT-L/14@336px w/ ST-Adapter
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ViT-L/14@336px w/ ST-AdapterPre-train data=CLIP+K4002022.06 | 98.3 | — | — | |
| ViT-L/14 w/ ST-AdapterPre-train data=CLIP+K4002022.06 | 98.1 | — | — | |
| CLAdapterBackbone=Swin-B2025.12 | 97.6 | — | — | |
| SlowOnly-8x8-R101Pre-train data=Kinetics+OmniSource[19]2022.06 | 97.3 | — | — | |
| FASTER32Pre-train data=K4002022.06 | 96.9 | — | — | |
| R(2+1)D-34Pre-train data=K4002022.06 | 96.8 | — | — | |
| S3DPre-train data=ImageNet+K4002022.06 | 96.8 | — | — | |
| VidPrism-MEvaluation Protocol=Tuning, Shot Count (K)=16, Backbone=VideoMAEv22026.05 | 96.6 | — | — | |
| ViT-B/16 w/ ST-AdapterPre-train data=CLIP+K4002022.06 | 96.4 | — | — | |
| VideoMAE2025.12 | 96.1 | — | — | |
| VidPrism-CEvaluation Protocol=Tuning, Shot Count (K)=16, Backbone=CLIP2026.05 | 96 | — | — | |
| VidPrism-MEvaluation Protocol=Tuning, Shot Count (K)=8, Backbone=VideoMAEv22026.05 | 95.9 | — | — | |
| STCPre-train data=K4002022.06 | 95.8 | — | — | |
| I3DPre-train data=ImageNet+K4002022.06 | 95.6 | — | — | |
| VidPrism-MEvaluation Protocol=Tuning, Shot Count (K)=4, Backbone=VideoMAEv22026.05 | 95.6 | — | — | |
| ETLEvaluation Protocol=Adapting, Shot Count (K)=162026.05 | 94.8 | — | — | |
| VideoMAEv2Evaluation Protocol=Tuning, Shot Count (K)=162026.05 | 94.7 | — | — | |
| CVRL2025.12 | 94.4 | — | — | |
| VidPrism-MEvaluation Protocol=Tuning, Shot Count (K)=2, Backbone=VideoMAEv22026.05 | 94.3 | — | — | |
| ρBYOLρ = 42025.12 | 94.2 | — | — | |
| OSTEvaluation Protocol=Tuning, Shot Count (K)=162026.05 | 93.9 | — | — | |
| VidPrism-CEvaluation Protocol=Tuning, Shot Count (K)=8, Backbone=CLIP2026.05 | 93.8 | — | — | |
| RSPNet2025.12 | 93.7 | — | — | |
| ECOPre-train data=K4002022.06 | 93.6 | — | — | |
| VideoPromptPre-train data=CLIP2022.06 | 93.6 | — | — | |
| MoTEEvaluation Protocol=Tuning, Shot Count (K)=162026.05 | 93.6 | — | — | |
| CORPf2025.12 | 93.5 | — | — | |
| MAXIEvaluation Protocol=Tuning, Shot Count (K)=162026.05 | 93.5 | — | — | |
| ViCLIPEvaluation Protocol=Tuning, Shot Count (K)=162026.05 | 93.2 | — | — | |
| ViFi-CLIPEvaluation Protocol=Adapting, Shot Count (K)=162026.05 | 92.7 | — | — | |
| ETLEvaluation Protocol=Adapting, Shot Count (K)=82026.05 | 92.7 | — | — | |
| MAXIEvaluation Protocol=Tuning, Shot Count (K)=82026.05 | 92.4 | — | — | |
| MoTEEvaluation Protocol=Tuning, Shot Count (K)=82026.05 | 92.3 | — | — | |
| VideoMAEv2Evaluation Protocol=Tuning, Shot Count (K)=82026.05 | 92.1 | — | — | |
| ViCLIPEvaluation Protocol=Tuning, Shot Count (K)=82026.05 | 92.1 | — | — | |
| VidPrism-CEvaluation Protocol=Tuning, Shot Count (K)=4, Backbone=CLIP2026.05 | 91.6 | — | — | |
| X-CLIPEvaluation Protocol=Adapting, Shot Count (K)=162026.05 | 91.4 | — | — | |
| ActionCLIPEvaluation Protocol=Adapting, Shot Count (K)=162026.05 | 91.4 | — | — | |
| OSTEvaluation Protocol=Tuning, Shot Count (K)=82026.05 | 91.3 | — | — | |
| MoTEEvaluation Protocol=Tuning, Shot Count (K)=42026.05 | 91 | — | — | |
| ETLEvaluation Protocol=Adapting, Shot Count (K)=42026.05 | 90.2 | — | — | |
| ViFi-CLIPEvaluation Protocol=Adapting, Shot Count (K)=82026.05 | 90 | — | — | |
| A5Evaluation Protocol=Adapting, Shot Count (K)=162026.05 | 89.9 | — | — | |
| MAXIEvaluation Protocol=Tuning, Shot Count (K)=42026.05 | 89.3 | — | — | |
| Vi²CLR2025.12 | 89.1 | — | — | |
| MoTEEvaluation Protocol=Tuning, Shot Count (K)=22026.05 | 88.8 | — | — | |
| OSTEvaluation Protocol=Tuning, Shot Count (K)=42026.05 | 88.4 | — | — | |
| X-CLIPEvaluation Protocol=Adapting, Shot Count (K)=82026.05 | 88.3 | — | — | |
| VidPrism-CEvaluation Protocol=Tuning, Shot Count (K)=2, Backbone=CLIP2026.05 | 88.1 | — | — | |
| ViCLIPEvaluation Protocol=Tuning, Shot Count (K)=42026.05 | 88 | — | — | |
| CoCLR2025.12 | 87.9 | — | — | |
| MAXIEvaluation Protocol=Tuning, Shot Count (K)=22026.05 | 86.8 | — | — | |
| VideoMAEv2Evaluation Protocol=Tuning, Shot Count (K)=42026.05 | 86.8 | — | — | |
| MemDPC2025.12 | 86.1 | — | — | |
| ETLEvaluation Protocol=Adapting, Shot Count (K)=22026.05 | 86.1 | — | — | |
| PCL (VCP)Pre-train=K400, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 85.7 | — | — | |
| A5Evaluation Protocol=Adapting, Shot Count (K)=82026.05 | 85.7 | — | — | |
| PCL (VCP)Pre-train=K400, ClipSize=16 x 112^2, Network=R3D-182020.10 | 85.6 | — | — | |
| ViFi-CLIPEvaluation Protocol=Adapting, Shot Count (K)=42026.05 | 85.1 | — | — | |
| PCL (VCP)Pre-train=UCF, ClipSize=16 x 112^2, Network=R3D-182020.10 | 83.4 | — | — | |
| X-CLIPEvaluation Protocol=Adapting, Shot Count (K)=42026.05 | 83.4 | — | — | |
| OSTEvaluation Protocol=Tuning, Shot Count (K)=22026.05 | 83 | — | — | |
| ViCLIPEvaluation Protocol=Tuning, Shot Count (K)=22026.05 | 83 | — | — | |
| PCL (3DRotNet)Pre-train=UCF, ClipSize=16 x 112^2, Network=R3D-182020.10 | 82.8 | — | — | |
| RTTPre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 81.6 | — | — | |
| PCL (VCP)Pre-train=UCF, ClipSize=16 x 112^2, Network=C3D2020.10 | 81.4 | — | — | |
| SpeedNetPre-train=K400, ClipSize=64 x 224^2, Network=S3D-G2020.10 | 81.1 | — | — | |
| PCL (VCP)Pre-train=UCF, ClipSize=16 x 112^2, Network=R3D2020.10 | 81.1 | — | — | |
| ViFi-CLIPEvaluation Protocol=Adapting, Shot Count (K)=22026.05 | 80.7 | — | — | |
| PCL (VCP)Pre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 79.9 | — | — | |
| A5Evaluation Protocol=Adapting, Shot Count (K)=42026.05 | 79.9 | — | — | |
| PCL (VCOP)Pre-train=UCF, ClipSize=16 x 112^2, Network=C3D2020.10 | 79.8 | — | — | |
| CBTPre-train=K600+, ClipSize=16 x 112^2, Network=S3D2020.10 | 79.5 | — | — | |
| RTTPre-train=K400, ClipSize=16 x 112^2, Network=R3D-182020.10 | 79.3 | — | — | |
| PCL (VCOP)Pre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 79.2 | — | — | |
| VideoMoCo2025.12 | 78.7 | — | — | |
| PCL (VCOP)Pre-train=UCF, ClipSize=16 x 112^2, Network=R3D2020.10 | 78.2 | — | — | |
| MemDPCPre-train=K400, ClipSize=40 x 224^2, Network=R-2D3D2020.10 | 78.1 | — | — | |
| RTTPre-train=UCF, ClipSize=16 x 112^2, Network=R3D-182020.10 | 77.3 | — | — | |
| PacePredPre-train=K400, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 77.1 | — | — | |
| X-CLIPEvaluation Protocol=Adapting, Shot Count (K)=22026.05 | 76.4 | — | — | |
| PacePredPre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 75.9 | — | — | |
| DPCPre-train=K400, ClipSize=16 x 224^2, Network=R3D-342020.10 | 75.7 | — | — | |
| IICPre-train=UCF, ClipSize=16 x 112^2, Network=R3D2020.10 | 74.4 | — | — | |
| VideoMAEv2Evaluation Protocol=Tuning, Shot Count (K)=22026.05 | 73.4 | — | — | |
| ActionCLIPEvaluation Protocol=Adapting, Shot Count (K)=82026.05 | 73 | — | — | |
| VCOPPre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 72.4 | — | — | |
| PRPPre-train=UCF, ClipSize=16 x 112^2, Network=R(2+1)D2020.10 | 72.1 | — | — | |
| ActionCLIPEvaluation Protocol=Adapting, Shot Count (K)=42026.05 | 71.5 | — | — | |
| A5Evaluation Protocol=Adapting, Shot Count (K)=22026.05 | 71.4 | — | — | |
| ViCLIPEvaluation Protocol=Zero-shot, Shot Count (K)=22026.05 | 71 | — | — | |
| ViCLIPEvaluation Protocol=Zero-shot, Shot Count (K)=42026.05 | 71 | — | — | |
| ViCLIPEvaluation Protocol=Zero-shot, Shot Count (K)=82026.05 | 71 | — | — | |
| ViCLIPEvaluation Protocol=Zero-shot, Shot Count (K)=162026.05 | 71 | — | — | |
| ActionCLIPEvaluation Protocol=Adapting, Shot Count (K)=22026.05 | 70 | — | — | |
| RTTPre-train=K400, ClipSize=16 x 112^2, Network=C3D2020.10 | 69.9 | — | — | |
| PRPPre-train=UCF, ClipSize=16 x 112^2, Network=C3D2020.10 | 69.1 | — | — | |
| VCPPre-train=UCF, ClipSize=16 x 112^2, Network=C3D2020.10 | 68.5 | — | — | |
| DPCPre-train=K400, ClipSize=16 x 128^2, Network=R3D-182020.10 | 68.2 | — | — | |
| PRPPre-train=UCF, ClipSize=16 x 112^2, Network=R3D2020.10 | 66.5 | — | — |