Video Instance Segmentation on OVIS (val)
57.1APCAVIS
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CAVISBackbone=ViT-L, temporal refiner=true2024.07 | 57.1 | 82.6 | 63.5 | 21.2 | 61.8 | — | — | — | |
| DVIS-DAQBackbone=ViT-L, Pre-training=DINOv2, GFLOPs=1173, FPS=8, Input resolution=5442026.02 | 54.3 | — | 60.2 | — | 59.8 | — | — | — | |
| DVIS++Backbone=ViT-L, temporal refiner=true2024.07 | 53.4 | 78.9 | 58.5 | 21.1 | 58.7 | — | — | — | |
| CAVISBackbone=ViT-L, temporal refiner=false2024.07 | 53.2 | 75.9 | 59.1 | 20.9 | 58.2 | — | — | — | |
| CAVISBackbone=ViT-L, Pre-training=DINOv2, GFLOPs=863, FPS=15, Input resolution=default2026.02 | 53.2 | — | 59.1 | — | 58.2 | — | — | — | |
| VidEoMTBackbone=ViT-L, Pre-training=DINOv2, GFLOPs=934, FPS=115, Input resolution=5442026.02 | 52.5 | — | 57.2 | — | 57.5 | — | — | — | |
| GLEE-ProBackbone=EVA02-L, Evaluation Protocol=Fine-tuned2023.12 | 50.4 | 71.4 | 55.5 | — | — | — | — | — | |
| DVISMode=Offline, Backbone=Swin-L, Resolution=720px2023.06 | 49.9 | 75.9 | 53 | 19.4 | 55.3 | — | — | — | |
| DVISBackbone=Swin-L2023.12 | 49.9 | 75.9 | 53 | — | — | — | — | — | |
| DVIS++Backbone=ViT-L, temporal refiner=false2024.07 | 49.6 | 72.5 | 55 | 20.8 | 54.6 | — | — | — | |
| DVIS++Backbone=ViT-L, Pre-training=DINOv2, GFLOPs=868, FPS=17, Input resolution=default2026.02 | 49.6 | — | 55 | — | 54.6 | — | — | — | |
| DVIS-DAQBackbone=Swin-L, Pre-training=IN21K, GFLOPs=423, FPS=12, Input resolution=default2026.02 | 49.5 | — | 51.7 | — | 54.9 | — | — | — | |
| UNINEXTBackbone=ViT-H, Online=true2023.03 | 49 | 72.5 | 52.2 | — | — | — | — | — | |
| UNINEXTBackbone=ViT-H2023.12 | 49 | 72.5 | 52.2 | — | — | — | — | — | |
| CAVISBackbone=Swin-L, temporal refiner=false2024.07 | 48.6 | 74 | 52.5 | 19.5 | 53.3 | — | — | — | |
| DVISMode=Offline, Backbone=Swin-L2023.06 | 48.6 | 74.7 | 50.5 | 18.8 | 53.8 | — | — | — | |
| DVISbackbone=Swin-L, processing_mode=offline2023.11 | 48.6 | 74.7 | 50.5 | 18.8 | 53.8 | — | — | — | |
| DVISMode=Online, Backbone=Swin-L, Resolution=720px2023.06 | 47.1 | 71.9 | 49.2 | 19.4 | 52.5 | — | — | — | |
| DVISbackbone=Swin-L, processing_mode=online/near-online2023.11 | 47.1 | 71.9 | 49.2 | 19.4 | 52.5 | — | — | — | |
| CTVISBackbone=Swin-L, temporal refiner=false2024.07 | 46.9 | 71.5 | 47.5 | 19.1 | 52.1 | — | — | — | |
| CTVISBackbone=Swin-L, Params.=216M2023.07 | 46.9 | 71.5 | 47.5 | 19.1 | 52.1 | — | — | — | |
| TCOVISBackbone=Swin-L, temporal refiner=false2024.07 | 46.7 | 70.9 | 49.5 | 19.1 | 50.8 | — | — | — | |
| RefineVISBackbone=Swin-L, Evaluation Protocol=online2023.06 | 46.1 | 69.7 | 47.8 | 19 | 50.8 | — | — | — | |
| RefineVISBackbone=Swin-L, Evaluation Protocol=offline2023.06 | 46 | 70.4 | 48.4 | 19.1 | 51.2 | — | — | — | |
| DVISBackbone=Swin-L, temporal refiner=false2024.07 | 45.9 | 71.1 | 48.3 | 18.5 | 51.5 | — | — | — | |
| DVISMode=Online, Backbone=Swin-L2023.06 | 45.9 | 71.1 | 48.3 | 18.5 | 51.5 | — | — | — | |
| DVISBackbone=Swin-L, Pre-training=IN21K, GFLOPs=419, FPS=24, Input resolution=default2026.02 | 45.9 | — | 48.3 | — | 51.5 | — | — | — | |
| GenVISBackbone=Swin-L, Evaluation Mode=semi-online2022.11 | 45.4 | 69.2 | 47.8 | 18.9 | 49 | — | — | — | |
| GenVISMode=Offline, Backbone=Swin-L, Use COCO pseudo videos=true2023.06 | 45.4 | 69.2 | 47.8 | 18.9 | 49 | — | — | — | |
| GenVISBackbone=Swin-L2023.12 | 45.4 | 69.2 | 47.8 | — | — | — | — | — | |
| GenVIS (near-online)backbone=Swin-L, processing_mode=near-online2023.11 | 45.4 | 69.2 | 47.8 | 18.9 | 49 | — | — | — | |
| GenVISBackbone=Swin-L, temporal refiner=false2024.07 | 45.2 | 69.1 | 48.4 | 19.1 | 48.6 | — | — | — | |
| GenVISBackbone=Swin-L, Evaluation Mode=online2022.11 | 45.2 | 69.1 | 48.4 | 19.1 | 48.6 | — | — | — | |
| GenVisMode=Online, Backbone=Swin-L, Use COCO pseudo videos=true2023.06 | 45.2 | 69.1 | 48.4 | 19.1 | 48.6 | — | — | — | |
| GenVIS (online)backbone=Swin-L, processing_mode=online2023.11 | 45.2 | 69.1 | 48.4 | 19.1 | 48.6 | — | — | — | |
| NOVISBackbone=Swin-L, T/S=4/2, MST=720, Note=Only evaluated but not trained on larger input resolution2023.08 | 43.5 | 68.3 | 43.8 | 19.4 | 46.9 | — | — | — | |
| NOVISBackbone=Swin-L2023.12 | 43.5 | 68.3 | 43.8 | — | — | — | — | — | |
| TarVISbackbone=Swin-L, processing_mode=online/near-online2023.11 | 43.2 | 67.8 | 44.6 | 18 | 50.4 | — | — | — | |
| NOVISBackbone=Swin-L, T/S=4/2, MST=3602023.08 | 43 | 66.9 | 44.5 | 18.9 | 46.3 | — | — | — | |
| MinVISBackbone=ViT-L, temporal refiner=false2024.07 | 42.9 | 65.7 | 45.4 | 19.8 | 46.5 | — | — | — | |
| IDOLBackbone=Swin-L2022.11 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| IDOLBackbone=Swin-L2023.06 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| IDOLSupervision=Pixel, Backbone=Swin Large (SwinL), Input Resolution=720p2023.03 | 42.6 | 65.7 | 45.2 | — | — | — | — | — | |
| MDQESupervision=Pixel, Backbone=Swin Large (SwinL), Input Resolution=720p2023.03 | 42.6 | 67.8 | 44.3 | — | — | 49.3 | 21.6 | 65.1 | |
| IDOLBackbone=Swin-L, Online=true2023.03 | 42.6 | 65.7 | 45.2 | — | — | — | — | — | |
| IDOLMode=Online, Backbone=Swin-L, Resolution=720px2023.06 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| ROVISMode=Online, Backbone=Swin-L, Resolution=720px2023.06 | 42.6 | 64.7 | 42.6 | 18.4 | 49.1 | — | — | — | |
| MDQEMode=Offline, Backbone=Swin-L, Resolution=720px2023.06 | 42.6 | 67.8 | 44.3 | 18.3 | 46.5 | — | — | — | |
| IDOLBackbone=Swin-L, Params.=213M2023.07 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| IDOLBackbone=Swin-L, T/S=ON, MST=7202023.08 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| IDOLBackbone=Swin-L2023.12 | 42.6 | 65.7 | 45.2 | — | — | — | — | — | |
| IDOLbackbone=Swin-L, processing_mode=online/near-online2023.11 | 42.6 | 65.7 | 45.2 | 17.9 | 49.6 | — | — | — | |
| MinVISSupervision=Pixel, Backbone=Swin Large (SwinL)2023.03 | 41.6 | 65.4 | 43.4 | — | — | 48.5 | 20.5 | 65.5 | |
| MinVISMode=Online, Backbone=Swin-L, Resolution=720px2023.06 | 41.6 | 65.2 | 42.8 | 19.3 | 45.1 | — | — | — | |
| ROVISMode=Online, Backbone=Swin-L2023.06 | 41.6 | 65 | 42.9 | 18.7 | 46.9 | — | — | — | |
| UNINEXTBackbone=ConvNeXt-L, Online=true2023.03 | 41.1 | 65.8 | 42 | — | — | — | — | — | |
| UNINEXTBackbone=ConvNeXt-L2023.12 | 41.1 | 65.8 | 42 | — | — | — | — | — | |
| BoxVISSupervision=Box, Backbone=Swin Large (SwinL), Tracker=clip-level tracker (MDQE)2023.03 | 40.6 | 68.4 | 39.9 | — | — | 45.8 | 20.9 | 59.4 | |
| GLEE-PlusBackbone=Swin-L, Evaluation Protocol=Fine-tuned2023.12 | 40.3 | 63.8 | 39.8 | — | — | — | — | — | |
| IDOLBackbone=Swin-L, temporal refiner=false2024.07 | 40 | 63.1 | 40.5 | 17.6 | 46.4 | — | — | — | |
| IDOLMode=Online, Backbone=Swin-L2023.06 | 40 | 63.1 | 40.5 | 17.6 | 46.4 | — | — | — | |
| Axial-VS w/ Tube-Linkbackbone=Swin-L, processing_mode=offline2023.11 | 39.8 | 64.5 | 40.1 | 17.9 | 43.7 | — | — | — | |
| MinVISBackbone=Swin-L, temporal refiner=false2024.07 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISBackbone=Swin-L2022.11 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISBackbone=Swin-L2023.06 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISMode=Online, Backbone=Swin-L2023.06 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISBackbone=Swin-L, Params.=216M2023.07 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISBackbone=Swin-L, T/S=ON, MST=3602023.08 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISbackbone=Swin-L, processing_mode=online/near-online2023.11 | 39.4 | 61.5 | 41.3 | 18.1 | 43.3 | — | — | — | |
| MinVISBackbone=Swin-L, Pre-training=IN21K, GFLOPs=408, FPS=30, Input resolution=default2026.02 | 39.4 | — | 41.3 | — | 43.3 | — | — | — | |
| Axial-VS w/ Tube-Linkbackbone=Swin-L, processing_mode=online/near-online2023.11 | 39.1 | 62.3 | 39.8 | 18.5 | 42.3 | — | — | — | |
| GLEE-ProBackbone=EVA02-L, Evaluation Protocol=Zero-shot2023.12 | 38.7 | 59.4 | 39.7 | — | — | — | — | — | |
| CAROQBackbone=Swin-L, temporal refiner=false2024.07 | 38.2 | 60.7 | 39.5 | 17.7 | 44.1 | — | — | — | |
| CAVISBackbone=ResNet-50, temporal refiner=false2024.07 | 37.6 | 63.4 | 38.2 | 16.5 | 43.5 | — | — | — | |
| VISAGEBackbone=ResNet-50, temporal refiner=false2024.07 | 36.2 | 60.3 | 35.3 | 16.1 | 40.3 | — | — | — | |
| GRAtt-VISTraining Protocol=video train2026.06 | 36.2 | 60.8 | 36.8 | 16.8 | 40 | — | — | — | |
| GenVISBackbone=ResNet-50, temporal refiner=false2024.07 | 35.8 | 60.8 | 36.2 | 16.3 | 39.6 | — | — | — | |
| GenVISBackbone=ResNet-50, Evaluation Mode=online2022.11 | 35.8 | 60.8 | 36.2 | 16.3 | 39.6 | — | — | — | |
| GenVIS (online)backbone=ResNet50, processing_mode=online2023.11 | 35.8 | 60.8 | 36.2 | 16.3 | 39.6 | — | — | — | |
| GenVISTraining Protocol=video train2026.06 | 35.8 | 60.8 | 36.2 | 16.3 | 39.6 | — | — | — | |
| CTVISBackbone=ResNet-50, temporal refiner=false2024.07 | 35.5 | 60.8 | 34.9 | 16.1 | 41.9 | — | — | — | |
| CTVISBackbone=ResNet-50, Params.=44M2023.07 | 35.5 | 60.8 | 34.9 | 16.1 | 41.9 | — | — | — | |
| CTVISTraining Protocol=video train2026.06 | 35.5 | 60.8 | 34.9 | 16.1 | 41.9 | — | — | — | |
| TCOVISBackbone=ResNet-50, temporal refiner=false2024.07 | 35.3 | 60.7 | 36.6 | 15.7 | 39.5 | — | — | — | |
| DeVISBackbone=SwinL, Method Mode=Near-online, T=6, S=42022.07 | 34.6 | 58.7 | 36.8 | — | — | — | — | — | |
| DeVISBackbone=Swin-L, T/S=6/4, MST=3602023.08 | 34.6 | 58.7 | 36.8 | — | — | — | — | — | |
| GenVISBackbone=ResNet-50, Evaluation Mode=semi-online2022.11 | 34.5 | 59.4 | 35 | 16.6 | 38.3 | — | — | — | |
| GenVISBackbone=ResNet-502023.12 | 34.5 | 59.4 | 35 | — | — | — | — | — | |
| GenVIS (near-online)backbone=ResNet50, processing_mode=near-online2023.11 | 34.5 | 59.4 | 35 | 16.6 | 38.3 | — | — | — | |
| BoxVISSupervision=Box, Backbone=Swin Large (SwinL)2023.03 | 34.4 | 59 | 33.8 | — | — | 39.5 | 14.6 | 56.8 | |
| DVISMode=Offline, Backbone=ResNet50, Resolution=720px2023.06 | 34.1 | 59.8 | 32.3 | 15.9 | 41.1 | — | — | — | |
| DVISBackbone=ResNet-502023.12 | 34.1 | 59.8 | 32.3 | — | — | — | — | — | |
| UNINEXTBackbone=ResNet-50, Online=true2023.03 | 34 | 55.5 | 35.6 | — | — | — | — | — | |
| UNINEXTBackbone=ResNet-502023.12 | 34 | 55.5 | 35.6 | — | — | — | — | — | |
| DVISMode=Offline, Backbone=ResNet502023.06 | 33.8 | 60.4 | 33.5 | 15.3 | 39.5 | — | — | — | |
| DVISbackbone=ResNet50, processing_mode=offline2023.11 | 33.8 | 60.4 | 33.5 | 15.3 | 39.5 | — | — | — | |
| RefineVISBackbone=ResNet-50, Evaluation Protocol=offline2023.06 | 33.7 | 56.2 | 34.8 | 15.6 | 39.8 | — | — | — | |
| RefineVISBackbone=ResNet-50, Evaluation Protocol=online2023.06 | 33.4 | 57.1 | 32.6 | 16.3 | 40.2 | — | — | — | |
| Tube-Linkbackbone=Swin-L, processing_mode=online/near-online, reproduced=true2023.11 | 33.3 | 54.6 | 32.8 | 16.8 | 37.7 | — | — | — | |
| MDQEType=720p, Backbone=ResNet50, FPS=13.5, Params=51.4M2023.03 | 33 | 57.4 | 32.2 | 15.4 | 38.4 | — | — | — |