Future Semantic Segmentation on CityScapes (test val)
46.25mIoU (All Classes)VFMF
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VFMFInitial Context Length |C|=4, Roll out future frames=6, Evaluation Protocol=Mean-of-k2025.12 | 46.25 | 43.66 | |
| DINO-ForesightInitial Context Length |C|=4, Roll out future frames=6, Evaluation Protocol=Deterministic2025.12 | 44.75 | 38.92 | |
| VFMFInitial Context Length |C|=4, Roll out future frames=6, Evaluation Protocol=Best-of-k2025.12 | 44.49 | 41.82 | |
| VFMFInitial Context Length |C|=3, Roll out future frames=7, Evaluation Protocol=Mean-of-k2025.12 | 44.31 | 41.64 | |
| VFMFInitial Context Length |C|=3, Roll out future frames=7, Evaluation Protocol=Best-of-k2025.12 | 42.41 | 39.66 | |
| DINO-ForesightInitial Context Length |C|=3, Roll out future frames=7, Evaluation Protocol=Deterministic2025.12 | 41.89 | 35.61 | |
| VFMFInitial Context Length |C|=2, Roll out future frames=8, Evaluation Protocol=Mean-of-k2025.12 | 41.69 | 38.92 | |
| VFMFInitial Context Length |C|=2, Roll out future frames=8, Evaluation Protocol=Best-of-k2025.12 | 39.57 | 36.7 | |
| DINO-ForesightInitial Context Length |C|=2, Roll out future frames=8, Evaluation Protocol=Deterministic2025.12 | 39.35 | 32.26 | |
| VFMFInitial Context Length |C|=1, Roll out future frames=9, Evaluation Protocol=Mean-of-k2025.12 | 34.05 | 30.94 | |
| VFMFInitial Context Length |C|=1, Roll out future frames=9, Evaluation Protocol=Best-of-k2025.12 | 31.74 | 28.56 | |
| DINO-ForesightInitial Context Length |C|=1, Roll out future frames=9, Evaluation Protocol=Deterministic2025.12 | 31.67 | 22 |