3D Visual Grounding on EmbodiedScan (Full)
62.18Overall AP@25DEGround
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DEGroundBackbone=ResNet-502025.06 | 62.18 | 62.76 | 55.7 | 63.56 | 61.4 | 63.02 | 65.98 | 52.95 | |
| BIP3DBackbone=Swin-T2025.06 | 54.66 | 55.07 | 50.12 | 55.78 | 54.03 | 61.23 | 55.41 | 39.36 | |
| DenseGroundingBackbone=ResNet-502025.06 | 44.41 | 45.31 | 34.23 | 44.42 | 44.4 | — | — | — | |
| ProxyTransformationBackbone=ResNet-502025.06 | 41.08 | 41.66 | 34.38 | 41.57 | 40.81 | — | — | — | |
| EmbodiedScanBackbone=ResNet-502025.06 | 39.41 | 40.12 | 31.45 | 40.21 | 38.96 | 41.99 | 41.53 | 30.29 | |
| L3Det2025.06 | 23.07 | 24.01 | 18.34 | 23.59 | 21.22 | — | — | — | |
| BUTD-DETR2025.06 | 22.14 | 23.12 | 18.23 | 22.47 | 20.98 | — | — | — | |
| ScanRefer2025.06 | 12.85 | 13.78 | 9.12 | 13.44 | 10.77 | — | — | — |