Gaze-following on VideoAttentionTarget (VAT) Inconsistent
0.9078AUCGazeLLE + VPT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GazeLLE + VPTBackbone=ViT-B, Adaptation protocol=Visual Prompt Tuning2026.05 | 0.9078 | 0.1576 | 20.56 | |
| HCLoRABackbone=ViT-L2026.05 | 0.9066 | 0.1459 | 18.83 | |
| HCLoRABackbone=ViT-B2026.05 | 0.9064 | 0.1513 | 18.34 | |
| GazeLLEBackbone=ViT-B2026.05 | 0.9059 | 0.167 | 24.21 | |
| GazeLLE + FTBackbone=ViT-B, Adaptation protocol=Full fine-tuning2026.05 | 0.9044 | 0.168 | 24.38 | |
| GazeLLE + LoRABackbone=ViT-B, Adaptation protocol=LoRA2026.05 | 0.9037 | 0.1637 | 22 | |
| Miao et al.2026.05 | 0.8954 | 0.158 | 16.45 | |
| Sharingan2026.05 | 0.8912 | 0.1509 | 16.34 | |
| Chong et al.2026.05 | 0.8551 | 0.1517 | 17.23 |