Instance Segmentation on COCO (APm, APm_50)
46.2APmGraphScan-Mamba-B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GraphScan-Mamba-BBackbone=GraphScan-Mamba-B, #P=112M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 46.2 | 70.3 | |
| GraphScan-Mamba-BBackbone=GraphScan-Mamba-B, #P=112M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 46 | 69.9 | |
| GraphScan-Mamba-SBackbone=GraphScan-Mamba-S, #P=69M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 45.9 | 69.9 | |
| GraphScan-Mamba-TBackbone=GraphScan-Mamba-T, #P=47M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 45.3 | 69.1 | |
| GraphScan-Mamba-SBackbone=GraphScan-Mamba-S, #P=69M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 45.3 | 69.1 | |
| DAMamba-BBackbone=DAMamba-B, #P=105M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 45.3 | 69.5 | |
| DAMamba-SBackbone=DAMamba-S, #P=65M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 45.1 | 69.2 | |
| Spatial-Mamba-BBackbone=Spatial-Mamba-B, #P=115M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 45.1 | 69.1 | |
| DAMamba-BBackbone=DAMamba-B, #P=105M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 44.9 | 68.9 | |
| DAMamba-TBackbone=DAMamba-T, #P=45M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 44.8 | 68.6 | |
| Spatial-Mamba-SBackbone=Spatial-Mamba-S, #P=63M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 44.6 | 68.7 | |
| DAMamba-SBackbone=DAMamba-S, #P=65M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 44.5 | 68.4 | |
| VMamba-SBackbone=VMamba-S, #P=70M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 44.2 | 68.2 | |
| LocalVMamba-SBackbone=LocalVMamba-S, #P=69M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 44.1 | 67.8 | |
| VMamba-BBackbone=VMamba-B, #P=108M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 44.1 | 68.3 | |
| Spatial-Mamba-SBackbone=Spatial-Mamba-S, #P=63M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 44 | 67.9 | |
| GraphScan-Mamba-TBackbone=GraphScan-Mamba-T, #P=47M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.9 | 67.7 | |
| Spatial-Mamba-TBackbone=Spatial-Mamba-T, #P=46M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.8 | 67.8 | |
| VMamba-TBackbone=VMamba-T, #P=50M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.7 | 67.4 | |
| BiFormer-BBackbone=BiFormer-B, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.7 | 67.6 | |
| VMamba-SBackbone=VMamba-S, #P=70M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.7 | 67.3 | |
| VSSD-TBackbone=VSSD-T, #P=44M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.6 | 67.6 | |
| VSSD-SBackbone=VSSD-S, #P=59M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.5 | 67.2 | |
| ConvNeXt-BBackbone=ConvNeXt-B, #P=107M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.5 | 67.1 | |
| LocalVMamba-TBackbone=LocalVMamba-T, #P=45M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.4 | 67 | |
| DAMamba-TBackbone=DAMamba-T, #P=45M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.4 | 67.2 | |
| Swin-BBackbone=Swin-B, #P=107M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.3 | 67.1 | |
| BiFormer-SBackbone=BiFormer-S, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.2 | 66.8 | |
| MSVMamba-TBackbone=MSVMamba-T, #P=53M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.2 | 66.8 | |
| Swin-SBackbone=Swin-S, #P=69M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.2 | 67 | |
| NAT-SBackbone=NAT-S, #P=70M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 43.2 | 66.9 | |
| LocalVMamba-SBackbone=LocalVMamba-S, #P=69M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 43.2 | 66.7 | |
| GroupMamba-TBackbone=GroupMamba-T, #P=40M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.9 | 66.5 | |
| Spatial-Mamba-TBackbone=Spatial-Mamba-T, #P=46M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.9 | 66.5 | |
| ConvNeXt-SBackbone=ConvNeXt-S, #P=70M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 42.9 | 66.9 | |
| VMamba-TBackbone=VMamba-T, #P=50M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.7 | 66.4 | |
| ConvNeXt-BBackbone=ConvNeXt-B, #P=107M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.7 | 66.3 | |
| NAT-TBackbone=NAT-T, #P=48M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 42.6 | 66.1 | |
| VSSD-TBackbone=VSSD-T, #P=44M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.6 | 66.4 | |
| Swin-BBackbone=Swin-B, #P=107M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.3 | 66 | |
| LocalVMamba-TBackbone=LocalVMamba-T, #P=45M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.2 | 65.7 | |
| MSVMamba-TBackbone=MSVMamba-T, #P=53M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 42.2 | 65.6 | |
| ConvNeXt-SBackbone=ConvNeXt-S, #P=70M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 41.8 | 65.2 | |
| ConvNeXt-TBackbone=ConvNeXt-T, #P=48M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 41.7 | 65 | |
| Swin-TBackbone=Swin-T, #P=48M, Schedule=3x+MS, Detector framework=Mask R-CNN2026.05 | 41.6 | 65.1 | |
| Swin-SBackbone=Swin-S, #P=69M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 40.9 | 65.3 | |
| ConvNeXt-TBackbone=ConvNeXt-T, #P=48M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 40.1 | 63.3 | |
| Swin-TBackbone=Swin-T, #P=48M, Schedule=1x, Detector framework=Mask R-CNN2026.05 | 39.3 | 62.2 |