Video Semantic Segmentation on VIPER to BDD 100k
84.4Road IoUHRDA + BLDA
Evaluation results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HRDA + BLDASegmentation network=Deeplabv2, Batch size=2, Image crop size=512 x 5122025.12 | 84.4 | 40.1 | 79.5 | 18.7 | 27.3 | 29.1 | 74.7 | 25.5 | 82.6 | 62.8 | 81.7 | 43.9 | 46.7 | 35.1 | 29.9 | 50.8 | 23.4 | |
| HRDASegmentation network=Deeplabv2, Batch size=2, Image crop size=512 x 5122025.12 | 84.2 | 35.5 | 80.7 | 13.1 | 24 | 28 | 75.1 | 25 | 84.1 | 60.4 | 84.4 | 37.8 | 44.7 | 30.6 | 29.7 | 49.2 | 25.2 |