Video Instance Segmentation on YouTube-VIS 2019
77.9APCAVIS
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| CAVISVar.=TCO, Backbone=ViT-L2026.06 | 77.9 | 98.7 | 87.8 | 63 | 81.3 | — | — | — | 11.5 | 8.3 | |
| CAVISVar.=TO, Backbone=ViT-L2026.06 | 73.8 | 94.4 | 82.4 | 59.7 | 77.6 | — | — | — | 7.3 | 4.9 | |
| CTVISVariant=TC-Oracle2026.06 | 73.5 | 97.7 | 82.1 | 59.6 | 76.7 | — | — | — | 22.2 | 15.7 | |
| VISAGEVariant=TC-Oracle2026.06 | 72.9 | 97.1 | 81.2 | 59.6 | 77 | — | — | — | 20.7 | 13.1 | |
| MinVISVariant=TC-Oracle2026.06 | 69.8 | 96.9 | 77 | 57.8 | 73.3 | — | — | — | 24.5 | 15.2 | |
| DVIS++Var.=TCO, Backbone=R502026.06 | 68.7 | 96.1 | 76.6 | 57.3 | 72.4 | — | — | — | 16.4 | 12 | |
| ViLLa2024.07 | 67.6 | — | — | — | — | — | — | — | — | — | |
| DVIS++Variant=TC-Oracle2026.06 | 67.1 | 93.9 | 72.9 | 56.3 | 70.2 | — | — | — | 14.9 | 10.7 | |
| UNINEXTBackbone=ViT-H, Joint Training=false, Universal=false2024.02 | 66.9 | — | — | — | — | — | — | — | — | — | |
| CAVISVariant=TC-Oracle2026.06 | 66.8 | 90.8 | 72.1 | 56.4 | 70.5 | — | — | — | 15.4 | 11.2 | |
| CAVISVar.=Def., Backbone=ViT-L2026.06 | 66.4 | 87.2 | 73.6 | 54.7 | 71 | — | — | — | — | — | |
| DVISVar.=TCO, Backbone=R502026.06 | 66 | 93.7 | 70.2 | 55.5 | 70.3 | — | — | — | 16.5 | 11.7 | |
| GenVISVariant=TC-Oracle2026.06 | 66 | 91.8 | 72 | 56 | 70.8 | — | — | — | 17.7 | 12.2 | |
| VISAGEVariant=T-Oracle2026.06 | 65.9 | 87.6 | 73.9 | 56.9 | 72.8 | — | — | — | 13.7 | 10.4 | |
| DVISVariant=TC-Oracle2026.06 | 65.9 | 94 | 71.5 | 55.5 | 69 | — | — | — | 18.1 | 12.7 | |
| GenVISVar.=TCO, Backbone=R502026.06 | 65.6 | 91.8 | 71.2 | 55.8 | 70.5 | — | — | — | 17.2 | 10.8 | |
| DVIS-offBackbone=SwinL, Joint Training=false, Universal=false2024.02 | 64.9 | — | — | — | — | — | — | — | — | — | |
| Tube-LinkBackbone=Swin-large2023.03 | 64.6 | 86.6 | 71.3 | 55.9 | 69.1 | — | — | — | — | — | |
| Tube-LinkBackbone=Swin-large2023.03 | 64.6 | — | — | — | — | — | — | — | — | — | |
| Tube-LinkBackbone=SwinL, Joint Training=false, Universal=false2024.02 | 64.6 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=Swin-large, COCO video pseudo labels=true2023.03 | 64.3 | 87.5 | 71 | 55.6 | 69.1 | — | — | — | — | — | |
| IDOLBackbone=Swin-large, COCO video pseudo labels=true2023.03 | 64.3 | — | — | — | — | — | — | — | — | — | |
| UNINEXTBackbone=ConvNeXtL, Joint Training=false, Universal=false2024.02 | 64.3 | — | — | — | — | — | — | — | — | — | |
| UNINEXTBackbone=ConvNeXt-L2024.11 | 64.3 | — | — | — | — | — | — | — | — | — | |
| IDOL2024.07 | 64.3 | — | — | — | — | — | — | — | — | — | |
| CTVISVariant=T-Oracle2026.06 | 64.3 | 87.8 | 70.3 | 53.6 | 69.2 | — | — | — | 13 | 9.7 | |
| Mask2FormerBackbone=InternVideo2_s1, #Params=6B2024.03 | 64.2 | — | — | — | — | — | — | — | — | — | |
| GenVIS-offBackbone=SwinL2024.02 | 63.8 | — | — | — | — | — | — | — | — | — | |
| HyperSeg-3BBackbone=Swin-B2024.11 | 63.8 | — | — | — | — | — | — | — | — | — | |
| Mask2FormerBackbone=InternViT (image), #Params=6B2024.03 | 63.4 | — | — | — | — | — | — | — | — | — | |
| VITABackbone=Swin-large, COCO video pseudo labels=true2023.03 | 63 | 86.9 | 67.9 | 56.3 | 68.1 | — | — | — | — | — | |
| VITABackbone=Swin-large, COCO video pseudo labels=true2023.03 | 63 | — | — | — | — | — | — | — | — | — | |
| VITA2024.07 | 63 | — | — | — | — | — | — | — | — | — | |
| CAVISVariant=T-Oracle2026.06 | 62.5 | 84.7 | 67.8 | 53.1 | 67.1 | — | — | — | 11.1 | 7.9 | |
| MinVISVariant=T-Oracle2026.06 | 62 | 87.1 | 68.1 | 52.6 | 67.2 | — | — | — | 16.7 | 10 | |
| Min-VISBackbone=Swin-large2023.03 | 61.6 | 83.3 | 68.6 | 54.8 | 66.6 | — | — | — | — | — | |
| Min-VISBackbone=Swin-large2023.03 | 61.6 | — | — | — | — | — | — | — | — | — | |
| Mask2Former2024.07 | 61.6 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=Swin-large2023.03 | 61.5 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=Swin-large2023.03 | 61.5 | — | — | — | — | — | — | — | — | — | |
| Mask2Former-VISBackbone=Swin-large2023.03 | 60.4 | 84.4 | 67 | — | — | — | — | — | — | — | |
| Mask2Former-VISBackbone=Swin-large2023.03 | 60.4 | — | — | — | — | — | — | — | — | — | |
| Mask2FormerBackbone=Swin-L (image), #Params=219M2024.03 | 60.3 | — | — | — | — | — | — | — | — | — | |
| UniVSBackbone=SwinL, Joint Training=true, Universal=true2024.02 | 60 | — | — | — | — | — | — | — | — | — | |
| DVIS++Var.=TO, Backbone=R502026.06 | 60 | 84 | 66.6 | 51.9 | 66.7 | — | — | — | 7.7 | 6.6 | |
| DVIS++Variant=T-Oracle2026.06 | 59.4 | 82.9 | 64.8 | 49.6 | 63.4 | — | — | — | 7.2 | 4 | |
| SeqFormerBackbone=Swin-large2023.03 | 59.3 | 82.1 | 66.4 | 51.7 | 64.4 | — | — | — | — | — | |
| SeqFormerBackbone=Swin-large2023.03 | 59.3 | — | — | — | — | — | — | — | — | — | |
| SeqFormer2024.07 | 59.3 | — | — | — | — | — | — | — | — | — | |
| UniVSBackbone=SwinB, Joint Training=true, Universal=true2024.02 | 57.8 | — | — | — | — | — | — | — | — | — | |
| DVISVariant=T-Oracle2026.06 | 57.7 | 80.9 | 62.7 | 50 | 62.3 | — | — | — | 9.9 | 7.2 | |
| TROY-VISBackbone=EfficientViT-L2, Fast Model=None (Upper Bound), FPS=7.42026.06 | 57.4 | 81.7 | 63.3 | — | — | — | — | — | — | — | |
| OMG-SegBackbone=ConvNeXt-L2024.11 | 56.4 | — | — | — | — | — | — | — | — | — | |
| MOBIUSBackbone=ResNet50, Fast Model=None (Upper Bound), FPS=8.02026.06 | 56.4 | 78.4 | 59.9 | — | — | — | — | — | — | — | |
| GenVISVar.=TO, Backbone=R502026.06 | 54.5 | 76.3 | 60 | 49.4 | 63.1 | — | — | — | 6.1 | 4.4 | |
| GenVISVariant=T-Oracle2026.06 | 53.3 | 72.8 | 59.5 | 49.8 | 63.1 | — | — | — | 5 | 6.1 | |
| UNINEXTBackbone=ResNet50, Joint Training=false, Universal=false2024.02 | 53 | — | — | — | — | — | — | — | — | — | |
| GLEEBackbone=ResNet50, Fast Model=None (Upper Bound), FPS=1.72026.06 | 52.9 | 73.8 | 58.9 | — | — | — | — | — | — | — | |
| Tube-LinkBackbone=ResNet502023.03 | 52.8 | 75.4 | 56.5 | 49.3 | 59.9 | — | — | — | — | — | |
| Tube-LinkBackbone=ResNet502023.03 | 52.8 | — | — | — | — | — | — | — | — | — | |
| Tube-LinkBackbone=ResNet50, Joint Training=false, Universal=false2024.02 | 52.8 | — | — | — | — | — | — | — | — | — | |
| DVIS-offBackbone=ResNet50, Joint Training=false, Universal=false2024.02 | 52.6 | — | — | — | — | — | — | — | — | — | |
| UniVSBackbone=SwinT, Joint Training=true, Universal=true2024.02 | 52.4 | — | — | — | — | — | — | — | — | — | |
| DVIS++Var.=Def., Backbone=R502026.06 | 52.3 | 75.6 | 57.7 | 45.3 | 57.7 | — | — | — | — | — | |
| VISAGEVariant=Default2026.06 | 52.2 | 74.6 | 58 | 46.4 | 57 | — | — | — | — | — | |
| DVIS++Variant=Default2026.06 | 52.2 | 75.6 | 56.2 | 45.6 | 56.9 | — | — | — | — | — | |
| VideoK-Net+Backbone=SwinB, Joint Training=false, Universal=false2024.02 | 51.4 | — | — | — | — | — | — | — | — | — | |
| CAVISVariant=Default2026.06 | 51.4 | 71.9 | 57 | 45.2 | 55.8 | — | — | — | — | — | |
| GenVISBackbone=ResNet502023.03 | 51.3 | 72 | 57.8 | 49.5 | 60 | — | — | — | — | — | |
| GenVIS-offBackbone=R502024.02 | 51.3 | — | — | — | — | — | — | — | — | — | |
| CTVISVariant=Default2026.06 | 51.3 | 73.5 | 55.7 | 43.9 | 54.9 | — | — | — | — | — | |
| MOBIUSBackbone=MNv4-CL, Fast Model=None (Upper Bound), FPS=8.62026.06 | 51 | 73 | 56.1 | — | — | — | — | — | — | — | |
| SegFSSlow Model=TROY-VIS, Backbone=EfficientViT-L2, FPS=23.32026.06 | 50.7 | 74.7 | 56 | — | — | — | — | — | — | — | |
| DVISVar.=TO, Backbone=R502026.06 | 50.1 | 70.7 | 54.1 | 48.4 | 61.9 | — | — | — | 0.6 | 4.6 | |
| VITABackbone=ResNet50, COCO video pseudo labels=true2023.03 | 49.8 | 72.6 | 54.5 | 49.4 | 61 | — | — | — | — | — | |
| VITABackbone=ResNet50, COCO video pseudo labels=true2023.03 | 49.8 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=ResNet50, COCO video pseudo labels=true2023.03 | 49.5 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=ResNet50, COCO video pseudo labels=true2023.03 | 49.5 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=R502024.02 | 49.5 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=SwinL2024.02 | 49.5 | — | — | — | — | — | — | — | — | — | |
| DVISVar.=Def., Backbone=R502026.06 | 49.5 | 71.8 | 54.9 | 43.8 | 55.3 | — | — | — | — | — | |
| MOBIUSBackbone=MNv4-CM, Fast Model=None (Upper Bound), FPS=8.72026.06 | 48.7 | 70.4 | 51.9 | — | — | — | — | — | — | — | |
| GenVISVar.=Def., Backbone=R502026.06 | 48.4 | 67.8 | 54.3 | 45 | 54.5 | — | — | — | — | — | |
| GenVISVariant=Default2026.06 | 48.3 | 67.6 | 53.3 | 43.8 | 53.4 | — | — | — | — | — | |
| DVISVariant=Default2026.06 | 47.8 | 69.4 | 53.3 | 42.8 | 53.9 | — | — | — | — | — | |
| TubeFormerBackbone=ResNet50, Axial Attention=true2023.03 | 47.5 | 68.7 | 52.1 | 50.2 | 59 | — | — | — | — | — | |
| TubeFormerBackbone=ResNet50 + Axial2023.03 | 47.5 | — | — | — | — | — | — | — | — | — | |
| TubeFormerBackbone=A-R50x64, Joint Training=false, Universal=false2024.02 | 47.5 | — | — | — | — | — | — | — | — | — | |
| SeqformerBackbone=ResNet502023.03 | 47.4 | 69.8 | 51.8 | 45.5 | 54.8 | — | — | — | — | — | |
| Min-VISBackbone=ResNet502023.03 | 47.4 | 69 | 52.1 | 45.7 | 55.7 | — | — | — | — | — | |
| SeqFormerBackbone=ResNet502023.03 | 47.4 | — | — | — | — | — | — | — | — | — | |
| Min-VISBackbone=ResNet502023.03 | 47.4 | — | — | — | — | — | — | — | — | — | |
| MinVISBackbone=R502024.02 | 47.4 | — | — | — | — | — | — | — | — | — | |
| UniVSBackbone=ResNet50, Joint Training=true, Universal=true2024.02 | 47.4 | — | — | — | — | — | — | — | — | — | |
| Mask2Former-VISBackbone=ResNet502023.03 | 46.4 | 68 | 50 | — | — | — | — | — | — | — | |
| IDOLBackbone=ResNet502023.03 | 46.4 | — | — | — | — | — | — | — | — | — | |
| Mask2Former-VISBackbone=ResNet502023.03 | 46.4 | — | — | — | — | — | — | — | — | — | |
| IDOLBackbone=ResNet502023.03 | 46.4 | — | — | — | — | — | — | — | — | — | |
| MinVISVariant=Default2026.06 | 45.3 | 66.7 | 49.9 | 42.6 | 50.7 | — | — | — | — | — | |
| SegFSSlow Model=MOBIUS, Backbone=ResNet50, FPS=30.22026.06 | 45 | 69.9 | 49 | — | — | — | — | — | — | — |