Action Recognition on IKEA-ASM (Top-1, Sym, N-Sym)
78.1Top-1 AccuracyFrame2Freq-ST
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Frame2Freq-STPretrain=DINOv2, Param (M)=3.5, Fine-tuning Protocol=PEFT2026.02 | 78.1 | 80.3 | 77 | |
| Frame2Freq-MSPretrain=CLIP, Param (M)=7.3, Fine-tuning Protocol=PEFT2026.02 | 78 | 83.9 | 75.1 | |
| Frame2Freq-MSPretrain=DINOv2, Param (M)=7.3, Fine-tuning Protocol=PEFT2026.02 | 76.4 | 81 | 74.2 | |
| M2-CLIPPretrain=DINOv2, Param (M)=14.8, Fine-tuning Protocol=PEFT2026.02 | 74.9 | 75.3 | 74.6 | |
| VideoSwin-BPretrain=K400, Param (M)=88, Fine-tuning Protocol=Fully Fine-tuned2026.02 | 72.6 | — | — | |
| VitaCLIPPretrain=DINOv2, Param (M)=29.0, Fine-tuning Protocol=PEFT2026.02 | 72.5 | 74.2 | 72.6 | |
| M2-CLIPPretrain=CLIP, Param (M)=14.8, Fine-tuning Protocol=PEFT2026.02 | 71.9 | 76.2 | 72.6 | |
| ST-AdapterPretrain=DINOv2, Param (M)=7.1, Fine-tuning Protocol=PEFT2026.02 | 70.5 | 68.5 | 73.9 | |
| ST-AdapterPretrain=CLIP, Param (M)=7.1, Fine-tuning Protocol=PEFT2026.02 | 70.2 | 70.3 | 70.7 | |
| Frame2Freq-STPretrain=CLIP, Param (M)=3.5, Fine-tuning Protocol=PEFT2026.02 | 70.1 | 75.6 | 66.1 | |
| VitaCLIPPretrain=CLIP, Param (M)=29.0, Fine-tuning Protocol=PEFT2026.02 | 63.3 | 73.1 | 58.6 |