Relative Camera Pose Estimation on Ego-Exo4D (val)
38.54Rotation Only AUC@5Ours
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OursInput Modality=Audio + Vision, Visual Condition=Clean2025.12 | 38.54 | 54.49 | 69.61 | 1.03 | 3.36 | 9.34 | 0.81 | 2.99 | 8.82 | |
| Ours w/o BinauralInput Modality=Audio + Vision, Visual Condition=Clean2025.12 | 37.99 | 53.87 | 69.04 | 0.93 | 3.12 | 8.8 | 0.74 | 2.8 | 8.34 | |
| Ours w/o DOAInput Modality=Audio + Vision, Visual Condition=Clean2025.12 | 37.48 | 53.45 | 68.88 | 0.96 | 3.21 | 8.92 | 0.75 | 2.86 | 8.41 | |
| Reloc3rInput Modality=Audio + Vision, Visual Condition=Clean2025.12 | 37.35 | 53.62 | 68.98 | 0.73 | 2.63 | 7.77 | 0.57 | 2.33 | 7.33 | |
| Reloc3r-AVInput Modality=Audio + Vision, Visual Condition=Clean2025.12 | 37.1 | 53.3 | 68.72 | 0.88 | 2.97 | 8.53 | 0.69 | 2.65 | 8.07 | |
| OursInput Modality=Audio + Vision, Visual Condition=Corrupted2025.12 | 27.17 | 40.66 | 54.46 | 0.49 | 1.77 | 5.45 | 0.35 | 1.44 | 4.8 | |
| Ours w/o BinauralInput Modality=Audio + Vision, Visual Condition=Corrupted2025.12 | 26.56 | 40.06 | 53.81 | 0.44 | 1.62 | 5.1 | 0.29 | 1.3 | 4.48 | |
| Reloc3r-AVInput Modality=Audio + Vision, Visual Condition=Corrupted2025.12 | 26.17 | 39.76 | 53.59 | 0.43 | 1.57 | 4.96 | 0.29 | 1.27 | 4.36 | |
| Ours w/o DOAInput Modality=Audio + Vision, Visual Condition=Corrupted2025.12 | 26.14 | 39.76 | 53.71 | 0.45 | 1.61 | 5.06 | 0.31 | 1.29 | 4.42 | |
| Reloc3rInput Modality=Audio + Vision, Visual Condition=Corrupted2025.12 | 25.17 | 38.88 | 52.8 | 0.34 | 1.28 | 4.28 | 0.21 | 1 | 3.66 | |
| DOA CP ModelInput Modality=Audio Only2025.12 | 19.07 | 33.37 | 50.07 | 0.09 | 0.37 | 1.38 | 0.03 | 0.14 | 0.78 | |
| ChanceInput Modality=Audio Only2025.12 | 19.02 | 33.33 | 50.03 | 0.06 | 0.26 | 1.11 | 0.02 | 0.12 | 0.76 | |
| NA-CP ModelInput Modality=Audio Only2025.12 | 18.83 | 33.27 | 50.05 | 0.07 | 0.29 | 1.16 | 0.02 | 0.12 | 0.72 |