Scene Classification on NYU Depth V2 (Modality-Specific)
50.8Accuracy (Fusion)LINet + prog. MD
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LINet + prog. MDPretrained=No2026.06 | 50.8 | 37.8 | 40.3 | |
| LINet (no MD)Pretrained=No2026.06 | 50.1 | 27 | 16.1 | |
| ResNet18 late fusion baselinePretrained=No2026.06 | 49.3 | — | — | |
| ResNet18 early fusion baselinePretrained=No2026.06 | 44.8 | — | — | |
| CoMAEPretrained=No2026.06 | 40.7 | — | — | |
| ResNet18 baselinePretrained=No2026.06 | — | 40.4 | 40.2 |