Embodied Referring Expression Comprehension on Refer360
37.05IoU@0.25ViLT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ViLTResidual Module Configuration=MuRes(V+L)2025.12 | 37.05 | 14.66 | |
| ViLTResidual Module Configuration=Without Residual2025.12 | 36.53 | 14.03 | |
| ViLTResidual Module Configuration=Vanilla Residual2025.12 | 35.34 | 14.37 | |
| Dual-EncoderResidual Module Configuration=MuRes(V)2025.12 | 31.36 | 8.92 | |
| Dual-EncoderResidual Module Configuration=Without Residual2025.12 | 31.08 | 9.83 | |
| Dual-EncoderResidual Module Configuration=MuRes(V+L)2025.12 | 31.08 | 10.68 | |
| Dual-EncoderResidual Module Configuration=Vanilla Residual2025.12 | 30.17 | 8.98 | |
| Dual-EncoderResidual Module Configuration=MuRes(L)2025.12 | 29.43 | 9.03 | |
| BLIP-2Residual Module Configuration=Without Residual2025.12 | 29.42 | 7.54 | |
| CLIPResidual Module Configuration=MuRes(V)2025.12 | 29.2 | 9.15 | |
| CLIPResidual Module Configuration=MuRes(L)2025.12 | 28.3 | 7.5 | |
| BLIP-2Residual Module Configuration=Vanilla Residual2025.12 | 27.66 | 7.31 | |
| CLIPResidual Module Configuration=Vanilla Residual2025.12 | 27.22 | 8.35 | |
| BLIP-2Residual Module Configuration=MuRes(L)2025.12 | 26.81 | 7.94 | |
| CLIPResidual Module Configuration=MuRes(V+L)2025.12 | 26.65 | 7.27 | |
| CLIPResidual Module Configuration=Without Residual2025.12 | 25.8 | 7.67 | |
| BLIP-2Residual Module Configuration=MuRes(V)2025.12 | 25.45 | 7.71 | |
| BLIP-2Residual Module Configuration=MuRes(V+L)2025.12 | 16.44 | 3.8 |