Instance Segmentation on Cityscapes v1 (val)
46.9APBPR+
Evaluation results
| Method | Links | |
|---|---|---|
| BPR+training data=fine + COCO, base model=PolyTransform2021.04 | 46.9 | |
| PolyTransformtraining data=fine + COCO2021.04 | 44.6 | |
| PANettraining data=fine + COCO2021.04 | 41.4 | |
| BPRtraining data=fine + COCO, base model=Mask R-CNN*2021.04 | 41.1 | |
| BPRtraining data=fine + COCO, base model=Mask R-CNN* + SegFix*2021.04 | 40.9 | |
| Mask R-CNN* + SegFix*training data=fine + COCO2021.04 | 38.2 | |
| UPSNettraining data=fine + COCO2021.04 | 37.8 | |
| SSAPtraining data=fine2021.04 | 37.3 | |
| Mask R-CNN*training data=fine + COCO2021.04 | 36.8 | |
| PANettraining data=fine2021.04 | 36.5 | |
| AdaptIStraining data=fine2021.04 | 36.3 | |
| BMask R-CNNtraining data=fine2021.04 | 35 | |
| Mask R-CNNtraining data=fine2021.04 | 31.5 | |
| SGNtraining data=fine + coarse2021.04 | 29.2 |