Panoptic Segmentation on COCO-C (2000 image subset)
47.2Mean PQPanoptic SegFormer
Evaluation results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Panoptic SegFormerBackbone=Swin-L2021.09 | 47.2 | 41.3 | 41.5 | 34.3 | 42.7 | 48.6 | 49.5 | 48.8 | 50.3 | 53.8 | 50.1 | 53.5 | 46.9 | 44.8 | 51.5 | 53.3 | 44.3 | |
| Panoptic SegFormerBackbone=PVTv2-B52021.09 | 47 | 41.5 | 41.1 | 36.1 | 42.5 | 48.4 | 49.6 | 48.4 | 50.4 | 53.5 | 50.8 | 53 | 46.2 | 42.4 | 50.3 | 52.9 | 44.3 | |
| MaskFormerBackbone=Swin-L2021.09 | 41.7 | 37.3 | 38 | 30.4 | 39.3 | 42.3 | 42.5 | 42.8 | 45.3 | 49.7 | 43.9 | 49.4 | 39.7 | 35.2 | 45.2 | 48.8 | 37.9 | |
| Panoptic SegFormerBackbone=ResNet-502021.09 | 32.9 | 26.9 | 30.2 | 17.5 | 31.6 | 35.5 | 27.9 | 35.4 | 38.6 | 45.7 | 31.3 | 43.9 | 29 | 24.3 | 35 | 41.9 | 32.3 | |
| D-DETRBackbone=ResNet-502021.09 | 30.3 | 25.6 | 28.7 | 16.8 | 29.7 | 32.5 | 24.9 | 31.4 | 35.9 | 43.1 | 28.6 | 41.3 | 24.5 | 21.7 | 31.7 | 39.7 | 28.7 | |
| MaskFormerBackbone=ResNet-502021.09 | 29.5 | 24.9 | 28.1 | 16.4 | 29.5 | 31.2 | 24.7 | 30.9 | 34.8 | 42.5 | 27.5 | 41.2 | 22 | 20.4 | 31 | 38.5 | 27.7 | |
| Panoptic FCNBackbone=ResNet-502021.09 | 26.8 | 22.5 | 23.7 | 14.1 | 25 | 28.2 | 20 | 28.3 | 31.9 | 39.4 | 24.3 | 38 | 22.9 | 20 | 29.6 | 35.3 | 25.3 |