Instance Segmentation on Cityscapes (test)
40.1AP (Overall)PolyTransform
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PolyTransformtraining data=fine+COCO2019.12 | 40.1 | 65.9 | — | — | 42.4 | 34.8 | 58.5 | 39.8 | 50 | 41.3 | 30.9 | 23.4 | — | — | |
| Axial-DeepLab-XLExtra Data=MV2020.03 | 39.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AInnoSegmentationExtra Data=COCO2019.11 | 39.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Panoptic-DeepLabExtra Data=MV2020.03 | 39 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NJUSTExtra Data=COCO2019.11 | 38.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Axial-DeepLab-LExtra Data=MV2020.03 | 38.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| iFLYTEK-CVExtra Data=COCO2019.11 | 38 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Sogou_MMExtra Data=COCO2019.11 | 37.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PANettraining data=fine + COCO2019.06 | 36.4 | 63.1 | — | — | 41.5 | 33.6 | 58.2 | 31.8 | 45.3 | 28.7 | 28.2 | 24.1 | — | — | |
| PANetExtra Data=COCO2019.11 | 36.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PANettraining data=fine+COCO2019.12 | 36.4 | 63.1 | — | — | 41.5 | 33.6 | 58.2 | 31.8 | 45.3 | 28.7 | 28.2 | 24.1 | — | — | |
| Mask R-CNNProtocol=Supervised, Specific Modules=ResNet, RPN, Examples Num=6, #Params=46M2024.03 | 36.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Panoptic-DeepLabExtra Data=None2020.03 | 34.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PolySnake†Training data=fine, FPS=4.22023.01 | 34.6 | 61.2 | — | — | 42.2 | 32.8 | 59.2 | 32 | 42.5 | 27.4 | 22.6 | 18.4 | — | — | |
| QueryInstBackbone=ResNet-502021.05 | 34.4 | 59.6 | — | — | 40.4 | 30.7 | 56.8 | 29.1 | 40.5 | 30.8 | 26 | 21.1 | — | — | |
| PolySnakeTraining data=fine, FPS=4.82023.01 | 34.3 | 61 | — | — | 41.3 | 31.8 | 58.4 | 31.9 | 42.4 | 28.6 | 22.4 | 17.9 | — | — | |
| Axial-DeepLab-XLExtra Data=None2020.03 | 34 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CondInstBackbone=DCN-101-BiFPN, semantic_branch=true2021.05 | 33.9 | 58.2 | — | — | 35.6 | 28.1 | 55 | 32.1 | 44.2 | 33.6 | 24.5 | 18.6 | — | — | |
| Axial-DeepLab-LExtra Data=None2020.03 | 33.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CondInstBackbone=ResNet-502021.05 | 33.2 | 57.2 | — | — | 35.1 | 27.7 | 54.5 | 29.5 | 42.3 | 33.8 | 23.9 | 18.9 | — | — | |
| UPSNetExtra Data=COCO2019.11 | 33 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UPSNettraining data=fine+COCO2019.12 | 33 | 59.7 | — | — | 35.9 | 27.4 | 51.9 | 31.8 | 43.1 | 31.4 | 23.8 | 19.1 | — | — | |
| UPSNetTraining data=Fine+COCO, backbone=ResNet50, fps=4.42022.03 | 33 | 59.6 | — | — | 35.9 | 27.4 | 51.8 | 31.7 | 43 | 31.3 | 23.7 | 19 | — | — | |
| UPSNetTraining data=fine + COCO, FPS=4.42023.01 | 33 | 59.6 | — | — | 35.9 | 27.4 | 51.8 | 31.7 | 43 | 31.3 | 23.7 | 19 | — | — | |
| UPSNetBackbone=ResNet-502021.05 | 33 | 59.7 | — | — | 35.9 | 27.4 | 51.9 | 31.8 | 43.1 | 31.4 | 23.8 | 19.1 | — | — | |
| BShapeNet+training data=fine+COCO2019.12 | 32.9 | 58.8 | — | — | 36.6 | 24.8 | 50.4 | 33.7 | 41 | 33.7 | 25.4 | 17.8 | — | — | |
| E2ECTraining data=Fine, backbone=DLA-34, fps=4.9, multi-component detection=true2022.03 | 32.9 | 59.2 | — | — | 39 | 27.8 | 56 | 29.5 | 41.2 | 29.1 | 21.3 | 19.6 | — | — | |
| E2ECTraining data=fine, FPS=4.92023.01 | 32.9 | 59.2 | — | — | 39 | 27.8 | 56 | 28.5 | 41.2 | 29.1 | 21.3 | 19.6 | — | — | |
| BShapeNet+Backbone=ResNet-502021.05 | 32.9 | 58.8 | — | — | 36.6 | 24.8 | 50.4 | 33.7 | 41 | 33.7 | 25.4 | 17.8 | — | — | |
| SSAPExtra Data=None2020.03 | 32.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSAPtraining data=fine2019.12 | 32.7 | 51.8 | — | — | 35.4 | 25.5 | 55.9 | 33.2 | 43.9 | 31.9 | 19.5 | 16.2 | — | — | |
| SSAPTraining data=fine2023.01 | 32.7 | 51.8 | — | — | 35.4 | 25.5 | 55.9 | 33.2 | 43.9 | 31.9 | 19.5 | 16.2 | — | — | |
| AdaptISExtra Data=None2019.11 | 32.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdaptIStraining data=fine2019.12 | 32.5 | 52.5 | — | — | 31.4 | 29.1 | 50 | 31.6 | 41.7 | 39.4 | 24.7 | 12.1 | — | — | |
| Mask R-CNNtraining data=fine + COCO2017.03 | 32 | 58.1 | — | — | 34.8 | 27 | 49.1 | 30.1 | 40.9 | 30.9 | 24.1 | 18.7 | — | — | |
| Mask R-CNNtraining data=fine+COCO2019.12 | 32 | 58.1 | — | — | 34.8 | 27 | 49.1 | 30.1 | 40.9 | 30.9 | 24.1 | 18.7 | — | — | |
| Mask R-CNNBackbone=ResNet-502021.05 | 32 | 58.1 | — | — | 34.8 | 27 | 49.1 | 30.1 | 40.9 | 30.9 | 24.1 | 18.7 | — | — | |
| Mask R-CNNtraining data=fine + COCO2019.06 | 31.9 | 58.1 | — | — | 34.8 | 27 | 49.1 | 30.1 | 40.9 | 30.9 | 24.1 | 18.7 | — | — | |
| PANettraining data=fine2019.06 | 31.8 | 57.1 | — | — | 36.8 | 30.4 | 54.8 | 27 | 36.3 | 25.5 | 22.6 | 20.8 | — | — | |
| PANetResolution=2048x1024, FPS=<12019.06 | 31.8 | 57.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| PANettraining data=fine2019.12 | 31.8 | 57.1 | — | — | 36.8 | 30.4 | 54.8 | 27 | 36.3 | 25.5 | 22.6 | 20.8 | — | — | |
| PANetTraining data=Fine, backbone=ResNet50, fps=<12022.03 | 31.8 | 57.1 | — | — | 36.8 | 30.4 | 54.8 | 27 | 36.3 | 25.5 | 22.6 | 20.8 | — | — | |
| PANetTraining data=fine, FPS=<12023.01 | 31.8 | 57.1 | — | — | 36.8 | 30.4 | 54.8 | 27 | 36.3 | 25.5 | 22.6 | 20.8 | — | — | |
| PANetMask type=Full2021.08 | 31.8 | 57.1 | — | 46 | — | — | — | — | — | — | — | — | 44.2 | — | |
| Deep SnakeTraining data=Fine, backbone=DLA-34, fps=4.6, multi-component detection=true2022.03 | 31.7 | 58.4 | — | — | 37.2 | 27 | 56 | 29.5 | 40.5 | 28.2 | 19 | 16.4 | — | — | |
| DeepSnakeTraining data=fine, FPS=4.62023.01 | 31.7 | 58.4 | — | — | 37.2 | 27 | 56 | 29.5 | 40.5 | 28.2 | 19 | 16.4 | — | — | |
| DANCETraining data=fine, FPS=6.32023.01 | 31.2 | 57.7 | — | — | 38.1 | 27.3 | 54 | 27.5 | 37.4 | 27.7 | 21.6 | 16.2 | — | — | |
| E2ECTraining data=Fine, backbone=DLA-34, fps=6.2, multi-component detection=false2022.03 | 30.3 | 54 | — | — | 40.7 | 27.9 | 55.4 | 28.4 | 35.8 | 20.1 | 20.9 | 13.2 | — | — | |
| Deep SnakeTraining data=Fine, backbone=DLA-342022.03 | 28.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GMIStraining data=fine + coarse2019.06 | 27.6 | 44.6 | — | — | 29.3 | 24.1 | 42.7 | 25.4 | 37.2 | 32.9 | 17.6 | 11.9 | — | — | |
| SpatialEmbeddings (ours)training data=fine2019.06 | 27.6 | 50.9 | — | — | 34.5 | 26.1 | 52.4 | 21.7 | 31.2 | 16.4 | 20.1 | 18.9 | — | — | |
| SpatialEmbeddingsResolution=2048x1024, FPS=112019.06 | 27.6 | 50.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Neven et al.training data=fine2019.12 | 27.6 | 50.9 | — | — | 34.5 | 26.1 | 52.4 | 21.7 | 31.2 | 16.4 | 20.1 | 18.9 | — | — | |
| GMISTraining data=Fine+coarse, backbone=ResNet1012022.03 | 27.6 | 49.6 | — | — | 29.3 | 24.1 | 42.7 | 25.4 | 37.2 | 32.9 | 17.6 | 11.9 | — | — | |
| SpatialTraining data=Fine, backbone=ERFNet, fps=112022.03 | 27.6 | 50.9 | — | — | 34.5 | 26.1 | 52.4 | 21.7 | 31.2 | 16.4 | 20.1 | 18.9 | — | — | |
| GMISTraining data=fine + coarse2023.01 | 27.6 | 49.6 | — | — | 29.3 | 24.1 | 42.7 | 25.4 | 37.2 | 32.9 | 17.6 | 11.9 | — | — | |
| SpatialTraining data=fine, FPS=112023.01 | 27.6 | 50.9 | — | — | 34.5 | 26.1 | 52.4 | 21.7 | 31.2 | 16.4 | 20.1 | 18.9 | — | — | |
| BShapeNet+training data=fine2019.12 | 27.3 | 50.4 | — | — | 29.7 | 23.4 | 46.7 | 26.1 | 33.3 | 24.8 | 20.3 | 14.1 | — | — | |
| GMIStraining data=fine+coarse2019.12 | 27.3 | 45.6 | — | — | 31.5 | 25.2 | 42.3 | 21.8 | 37.2 | 28.9 | 18.8 | 12.8 | — | — | |
| Mask R-CNNMask type=Full, Runtime(s)=0.22021.08 | 26.22 | 49.89 | — | 40.11 | — | — | — | — | — | — | — | — | 37.63 | — | |
| Mask R-CNN2017.08 | 26.2 | 49.9 | — | 40.1 | — | — | — | — | — | — | — | — | 37.6 | — | |
| Mask R-CNNtraining data=fine2017.03 | 26.2 | 49.9 | — | — | 30.5 | 23.7 | 46.9 | 22.8 | 32.2 | 18.6 | 19.1 | 16 | — | — | |
| Mask R-CNNtraining data=fine2019.06 | 26.2 | 49.9 | — | — | 30.5 | 23.7 | 46.9 | 22.8 | 32.2 | 18.6 | 19.1 | 16 | — | — | |
| Mask RCNNResolution=2048x1024, FPS=2.2, mode=fine2019.06 | 26.2 | 49.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mask R-CNNtraining data=fine2019.12 | 26.2 | 49.9 | — | — | 30.5 | 23.7 | 46.9 | 22.8 | 32.2 | 18.6 | 19.1 | 16 | — | — | |
| Mask R-CNNTraining data=Fine, backbone=ResNet50, fps=2.22022.03 | 26.2 | 49.9 | — | — | 30.5 | 23.7 | 46.9 | 22.8 | 32.2 | 18.6 | 19.1 | 16 | — | — | |
| Mask R-CNNTraining data=fine, FPS=2.22023.01 | 26.2 | 49.9 | — | — | 30.5 | 23.7 | 46.9 | 22.8 | 32.2 | 18.6 | 19.1 | 16 | — | — | |
| PolygonRNN++training data=fine2019.06 | 25.5 | 45.5 | — | — | 29.4 | 21.8 | 48.3 | 21.1 | 32.3 | 23.7 | 13.6 | 13.6 | — | — | |
| PolygonRNN++training data=fine2019.12 | 25.5 | 45.5 | — | — | 29.4 | 21.8 | 48.3 | 21.2 | 32.3 | 23.7 | 13.6 | 13.6 | — | — | |
| Polygon RNN++Training data=Fine, backbone=ResNet502022.03 | 25.5 | 45.5 | — | — | 29.4 | 21.8 | 48.3 | 21.1 | 32.3 | 23.7 | 13.6 | 13.6 | — | — | |
| PolygonRNN++Training data=fine2023.01 | 25.5 | 45.5 | — | — | 29.4 | 21.8 | 48.3 | 21.1 | 32.3 | 23.7 | 13.6 | 13.6 | — | — | |
| SGNtraining data=fine + coarse2017.03 | 25 | 44.9 | — | — | 21.8 | 20.1 | 39.4 | 24.8 | 33.2 | 30.8 | 17.7 | 12.4 | — | — | |
| SGNapproach=proposal-free2018.03 | 25 | 44.9 | — | 44.5 | — | — | — | — | — | — | — | — | 38.9 | — | |
| SGNtraining data=fine + coarse2019.06 | 25 | 44.9 | — | — | 21.8 | 20.1 | 39.4 | 24.8 | 33.2 | 30.8 | 17.7 | 12.4 | — | — | |
| SGNResolution=2048x1024, FPS=0.62019.06 | 25 | 44.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SGNtraining data=fine+coarse2019.12 | 25 | 44.9 | — | — | 21.8 | 20.1 | 39.4 | 24.8 | 33.2 | 30.8 | 17.7 | 12.4 | — | — | |
| SGNTraining data=Fine+coarse, backbone=VGG16, fps=0.62022.03 | 25 | 44.9 | — | — | 21.8 | 20.1 | 39.4 | 24.8 | 33.2 | 30.8 | 17.7 | 12.4 | — | — | |
| SGNTraining data=fine + coarse, FPS=0.62023.01 | 25 | 44.9 | — | — | 21.8 | 20.1 | 39.4 | 24.8 | 33.2 | 30.8 | 17.7 | 12.4 | — | — | |
| DINtraining data=fine + coarse2019.06 | 23.4 | 45.2 | — | — | 20.9 | 18.4 | 31.7 | 22.8 | 31.1 | 31 | 19.6 | 11.7 | — | — | |
| Arnab et al.training data=fine2019.12 | 23.4 | 45.2 | — | — | 21 | 18.4 | 31.7 | 22.8 | 31.1 | 31 | 19.6 | 11.7 | — | — | |
| Kendall et al.training data=fine2019.12 | 21.6 | 39 | — | — | 19.2 | 21.4 | 36.6 | 18.8 | 26.8 | 15.9 | 19.4 | 14.5 | — | — | |
| GiT-LTraining=universal, Specific Modules=None, Examples Num=1, #Params=387M2024.03 | 20.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Pixelwise DIN2017.08 | 20 | 38.8 | — | 37.6 | — | — | — | — | — | — | — | — | 32.6 | — | |
| DINtraining data=fine + coarse2017.03 | 20 | 38.8 | — | — | 16.5 | 16.7 | 25.7 | 20.6 | 30 | 23.4 | 17.1 | 10.1 | — | — | |
| Dynamic NetResolution=2048x1024, FPS=<32019.06 | 20 | 38.3 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deep Watershed Transform2016.11 | 19.4 | 35.3 | 31.4 | 36.8 | — | — | — | — | — | — | — | — | — | — | |
| DWT2017.08 | 19.4 | 35.3 | — | 36.8 | — | — | — | — | — | — | — | — | 31.4 | — | |
| DWTapproach=proposal-free2018.03 | 19.4 | 35.3 | — | 36.8 | — | — | — | — | — | — | — | — | 31.4 | — | |
| DWTResolution=2048x1024, FPS=<32019.06 | 19.4 | 35.3 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DWTtraining data=fine2019.12 | 19.4 | 35.3 | — | — | 15.5 | 14.1 | 31.5 | 22.5 | 27 | 22.9 | 13.9 | 8 | — | — | |
| GiT-HTraining=universal, Specific Modules=None, Examples Num=1, #Params=756M2024.03 | 18.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GiT-BTraining=universal, Specific Modules=None, Examples Num=1, #Params=131M2024.03 | 17.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| discriminative loss function2017.08 | 17.5 | 35.9 | — | 31 | — | — | — | — | — | — | — | — | 27.8 | — | |
| DLapproach=proposal-free2018.03 | 17.5 | 35.9 | — | 31 | — | — | — | — | — | — | — | — | 27.8 | — | |
| Discriminate LossResolution=2048x1024, FPS=52019.06 | 17.5 | 35.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Boundary-aware2017.08 | 17.4 | 36.7 | — | 34 | — | — | — | — | — | — | — | — | 29.3 | — | |
| SAIStraining data=fine2017.03 | 17.4 | 36.7 | — | — | 14.6 | 12.9 | 35.7 | 16 | 23.2 | 19 | 10.3 | 7.8 | — | — | |
| BAISResolution=2048x1024, FPS=<12019.06 | 17.4 | 36.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DWTtraining data=fine2017.03 | 15.6 | 30 | — | — | 15.1 | 11.7 | 32.9 | 17.1 | 20.4 | 15 | 7.9 | 4.9 | — | — | |
| CenterPolyMask type=Polygon, Runtime(s)=0.0462021.08 | 15.54 | 39.49 | — | 24.45 | — | — | — | — | — | — | — | — | 23.33 | — |