We propose Global Aggregation then Local Distribution (GALD) scheme to distribute global information to each position adaptively according to the local information around the position. GALD net achieves top performance on Cityscapes dataset. Both source code and models will be available soon.
| Method | Conference | Backbone | mIoU(%) |
|---|---|---|---|
| RefineNet | CVPR2017 | ResNet-101 | 73.6 |
| SAC | ICCV2017 | ResNet-101 | 78.1 |
| PSPNet | CVPR2017 | ResNet-101 | 78.4 |
| DUC-HDC | WACV2018 | ResNet-101 | 77.6 |
| AAF | ECCV2018 | ResNet-101 | 77.1 |
| BiSeNet | ECCV2018 | ResNet-101 | 78.9 |
| PSANet | ECCV2018 | ResNet-101 | 80.1 |
| DFN | CVPR2018 | ResNet-101 | 79.3 |
| DSSPN | CVPR2018 | ResNet-101 | 77.8 |
| DenseASPP | CVPR2018 | DenseNet-161 | 80.6 |
| OCNet | - | ResNet-101 | 81.7 |
| CCNet | - | ResNet-101 | 81.4 |
| GALD-Net | - | ResNet50 | 80.8 |
| GALD-Net | - | ResNet101 | 81.8 |
| GALD-Net(use coarse data) | - | ResNet101 | 82.9 |
| GALD-Net(use Mapillary) | - | ResNet101 | 83.3 |
Detailed Results are shown here (Single Model Result)