Action Recognition on IKEA-ASM (Sym Acc., N-Sym Acc., Ovr. Acc.)
76.8Ovr. Acc.STEP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| STEPBackbone=DINOv2, GFLOPs=413.1, Train Params (M)=2.62025.03 | 76.8 | 69.46 | 80.59 | |
| M2-CLIPBackbone=DINOv2, GFLOPs=1128, Train Params (M)=14.82025.03 | 74.88 | 68.49 | 77.95 | |
| VideoSWINv2Backbone=IN-21K, GFLOPs=282, Train Params (M)=88.12025.03 | 72.6 | — | — | |
| VitaCLIPBackbone=DINOv2, GFLOPs=1130, Train Params (M)=28.62025.03 | 72.48 | 66.03 | 74.97 | |
| M2-CLIPBackbone=CLIP, GFLOPs=935, Train Params (M)=14.82025.03 | 71.86 | 64.11 | 76.62 | |
| ST-AdaptorBackbone=DINOv2, GFLOPs=1099, Train Params (M)=7.12025.03 | 70.54 | 63.48 | 73.94 | |
| ST-AdaptorBackbone=CLIP, GFLOPs=911, Train Params (M)=7.12025.03 | 70.15 | 54.17 | 77.84 | |
| Attn ProbingBackbone=DINOv2, GFLOPs=356.3, Train Params (M)=7.32025.03 | 65.76 | 64.73 | 70.64 | |
| STEPBackbone=CLIP, GFLOPs=307.8, Train Params (M)=2.62025.03 | 64.83 | 64.65 | 70.44 | |
| VitaCLIPBackbone=CLIP, GFLOPs=937, Train Params (M)=28.62025.03 | 63.25 | 50.11 | 69.57 | |
| Attn ProbingBackbone=CLIP, GFLOPs=273.6, Train Params (M)=7.32025.03 | 61.94 | 60.77 | 64.43 | |
| Self-Attn ProbingBackbone=CLIP, GFLOPs=308.1, Train Params (M)=2.62025.03 | 61.39 | 59.35 | 65.67 | |
| Self-Attn ProbingBackbone=DINOv2, GFLOPs=413.5, Train Params (M)=2.62025.03 | 60.78 | 56.3 | 68.01 | |
| Linear ProbingBackbone=DINOv2, GFLOPs=351.5, Train Params (M)=0.32025.03 | 41.71 | 32.45 | 46.15 | |
| Linear ProbingBackbone=CLIP, GFLOPs=269.6, Train Params (M)=0.32025.03 | 32.7 | 25.29 | 36.39 |