Video Object Segmentation on DAVIS 2016 (test)
86.6mIoULucidTracker
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| LucidTracker# training images=24-126, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 86.6 | — | — | — | — | — | — | — | — | — | |
| VideoGCRF# training images=~120k, Flow (F)=false, Annotation Protocol=Uses 1st frame annotation2017.03 | 86.5 | — | — | — | — | — | — | — | — | — | |
| OnAVOS# training images=~120k, Flow (F)=false, Annotation Protocol=Uses 1st frame annotation2017.03 | 86.1 | — | — | — | — | — | — | — | — | — | |
| PREMVOS# training images=~145k, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 84.9 | — | — | — | — | — | — | — | — | — | |
| MaskTrack# training images=~11k, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 80.3 | — | — | — | — | — | — | — | — | — | |
| OSVOS# training images=~2.3k, Flow (F)=false, Annotation Protocol=Uses 1st frame annotation2017.03 | 79.8 | — | — | — | — | — | — | — | — | — | |
| PDB# training images=~18k, Flow (F)=false, Annotation Protocol=Ignores 1st frame annotation2017.03 | 77.2 | — | — | — | — | — | — | — | — | — | |
| LVO# training images=~35k, Flow (F)=true, Annotation Protocol=Ignores 1st frame annotation2017.03 | 75.9 | — | — | — | — | — | — | — | — | — | |
| VPN# training images=~2.3k, Flow (F)=false, Annotation Protocol=Uses 1st frame annotation2017.03 | 75 | — | — | — | — | — | — | — | — | — | |
| MaskTrackInput=First-frame segmentation mask2016.12 | 74.8 | — | — | — | — | — | — | — | — | — | |
| MaskTrack_BoxInput=First-frame box annotation2016.12 | 73.7 | — | — | — | — | — | — | — | — | — | |
| STV# training images=0, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 73.6 | — | — | — | — | — | — | — | — | — | |
| FusionSeg# training images=~95k, Flow (F)=true, Annotation Protocol=Ignores 1st frame annotation2017.03 | 71.5 | — | — | — | — | — | — | — | — | — | |
| ObjFlow2016.12 | 71.4 | — | — | — | — | — | — | — | — | — | |
| ObjFlow# training images=0, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 71.1 | — | — | — | — | — | — | — | — | — | |
| Flow saliency# training images=0, Flow (F)=true, Annotation Protocol=Ignores 1st frame annotation2017.03 | 70.7 | — | — | — | — | — | — | — | — | — | |
| MP-Net# training images=~22.5k, Flow (F)=true, Annotation Protocol=Ignores 1st frame annotation2017.03 | 69.7 | — | — | — | — | — | — | — | — | — | |
| Grabcut oracleInput=Segmentation from Grabcut2016.12 | 67.3 | — | — | — | — | — | — | — | — | — | |
| Grabcut oracle# training images=0, Flow (F)=false, Annotation Protocol=Ignores 1st frame annotation2017.03 | 67.3 | — | — | — | — | — | — | — | — | — | |
| BVS2016.12 | 66.5 | — | — | — | — | — | — | — | — | — | |
| BVS# training images=0, Flow (F)=false, Annotation Protocol=Uses 1st frame annotation2017.03 | 66.5 | — | — | — | — | — | — | — | — | — | |
| NLC2016.12 | 64.1 | — | — | — | — | — | — | — | — | — | |
| NLC# training images=0, Flow (F)=true, Annotation Protocol=Ignores 1st frame annotation2017.03 | 64.1 | — | — | — | — | — | — | — | — | — | |
| FCP2016.12 | 63.1 | — | — | — | — | — | — | — | — | — | |
| FCP# training images=0, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 63.1 | — | — | — | — | — | — | — | — | — | |
| Box oracleInput=Box annotation2016.12 | 45.1 | — | — | — | — | — | — | — | — | — | |
| Box oracle# training images=0, Flow (F)=false, Annotation Protocol=Ignores 1st frame annotation2017.03 | 45.1 | — | — | — | — | — | — | — | — | — | |
| Saliency# training images=0, Flow (F)=false, Annotation Protocol=Ignores 1st frame annotation2017.03 | 32.7 | — | — | — | — | — | — | — | — | — | |
| Mask warping# training images=0, Flow (F)=true, Annotation Protocol=Uses 1st frame annotation2017.03 | 32.1 | — | — | — | — | — | — | — | — | — | |
| [19]2022.03 | — | 65.3 | 77.3 | — | 65.1 | 74.1 | — | — | — | — | |
| 3DC-Seg2022.09 | — | — | — | — | — | — | — | — | 84.3 | 84.7 | |
| 3DC-SegSupervision=true, RGB=true, Flow=true, Resolution=854 x 480, Runtime=0.84s2021.04 | — | 84.3 | — | — | — | — | — | — | — | — | |
| AMDFully unsupervised=true2022.09 | — | — | — | — | — | — | — | — | 57.8 | — | |
| ARPLearning Protocol=Unsupervised2019.07 | — | 76.2 | 91.1 | 7 | 70.6 | 83.5 | 7.9 | 39.3 | — | — | |
| ARPOptical flow=CPMFlow [32]2022.09 | — | — | — | — | — | — | — | — | 76.2 | 70.6 | |
| ARPSupervision=true, RGB=true, Flow=true, Runtime=74.5s2021.04 | — | 76.2 | — | — | — | — | — | — | — | — | |
| BATMANOptical flow=RAFT [73]2022.09 | — | — | — | — | — | — | — | — | 90.7 | 94.2 | |
| CISTraining on videos=true, Optical flow=PWCNet [71]2022.09 | — | — | — | — | — | — | — | — | 71.5 | 70.5 | |
| CISSupervision=false, RGB=true, Flow=true, Resolution=192 x 384, Runtime=0.1s, Post-processing=false2021.04 | — | 59.2 | — | — | — | — | — | — | — | — | |
| CISSupervision=false, RGB=true, Flow=true, Resolution=192 x 384, Runtime=11s, Post-processing=true2021.04 | — | 71.5 | — | — | — | — | — | — | — | — | |
| COSNetSupervision=true, RGB=true, Flow=false, Resolution=473 x 4732021.04 | — | 80.5 | — | — | — | — | — | — | — | — | |
| CUTLearning Protocol=Unsupervised2019.07 | — | 55.2 | 57.5 | 2.2 | 55.2 | 61 | 3.4 | 27.7 | — | — | |
| CUTOptical flow=LDOF [3], Fully unsupervised=true2022.09 | — | — | — | — | — | — | — | — | 55.2 | 55.2 | |
| CUTSupervision=false, RGB=true, Flow=true, Runtime=103s2021.04 | — | 55.2 | — | — | — | — | — | — | — | — | |
| DSTraining on videos=true, Optical flow=RAFT [73]2022.09 | — | — | — | — | — | — | — | — | 79.1 | — | |
| DyStaB (Supervised)Training on videos=true, Optical flow=PWCNet [71]2022.09 | — | — | — | — | — | — | — | — | 82.8 | — | |
| DyStaB (Unsupervised)Optical flow=PWCNet [71]2022.09 | — | — | — | — | — | — | — | — | 80 | — | |
| FSEGLearning Protocol=Unsupervised2019.07 | — | 70.7 | 83.5 | 1.5 | 65.3 | 73.8 | 1.8 | 32.8 | — | — | |
| FSEGTraining on videos=true2022.09 | — | — | — | — | — | — | — | — | 70.7 | 65.3 | |
| FSEGSupervision=true, RGB=true, Flow=true, Resolution=854 x 4802021.04 | — | 70.7 | — | — | — | — | — | — | — | — | |
| FSTLearning Protocol=Unsupervised2019.07 | — | 55.8 | 64.9 | 0 | 51.1 | 51.6 | 2.9 | 36.6 | — | — | |
| FTSOptical flow=LDOF [72], Fully unsupervised=true2022.09 | — | — | — | — | — | — | — | — | 55.8 | 51.1 | |
| FTSSupervision=false, RGB=true, Flow=true, Runtime=0.5s2021.04 | — | 55.8 | — | — | — | — | — | — | — | — | |
| LMPLearning Protocol=Unsupervised2019.07 | — | 70 | 85 | 1.3 | 65.9 | 79.2 | 2.5 | 57.2 | — | — | |
| LSMOLearning Protocol=Unsupervised2019.07 | — | 78.2 | 89.1 | 4.1 | 75.9 | 84.7 | 3.5 | 21.2 | — | — | |
| LVOTraining on videos=true, Optical flow=MP-Net [74]2022.09 | — | — | — | — | — | — | — | — | 75.9 | 72.1 | |
| LVOSupervision=true, RGB=true, Flow=true2021.04 | — | 75.9 | — | — | — | — | — | — | — | — | |
| MATNetTraining on videos=true, Optical flow=PWCNet [71]2022.09 | — | — | — | — | — | — | — | — | 82.4 | 80.7 | |
| MATNetSupervision=true, RGB=true, Flow=true, Resolution=473 x 473, Runtime=0.55s2021.04 | — | 82.4 | — | — | — | — | — | — | — | — | |
| MoSegTraining on videos=true, Optical flow=RAFT [73]2022.09 | — | — | — | — | — | — | — | — | 68.3 | 61.1 | |
| MotAdaptLearning Protocol=Unsupervised2019.07 | — | 77.2 | 87.8 | 5 | 77.4 | 84.4 | 3.3 | 27.9 | — | — | |
| Motion GroupingSupervision=false, RGB=false, Flow=true, Resolution=128 x 224, Runtime=0.012s2021.04 | — | 68.3 | — | — | — | — | — | — | — | — | |
| MSgStP2022.09 | — | — | — | — | — | — | — | — | 77.6 | — | |
| NLCLearning Protocol=Unsupervised2019.07 | — | 55.1 | 55.8 | 12.6 | 52.3 | 51.9 | 11.4 | 42.5 | — | — | |
| NLCOptical flow=SIFTFlow [45]2022.09 | — | — | — | — | — | — | — | — | 55.1 | 52.3 | |
| NLCSupervision=false, RGB=true, Flow=true, Runtime=11s2021.04 | — | 55.1 | — | — | — | — | — | — | — | — | |
| OursTraining on videos=true, Optical flow=ARFlow [46], Fully unsupervised=true2022.09 | — | — | — | — | — | — | — | — | 80.2 | 77.5 | |
| Ours (3D CNN)Learning Protocol=Unsupervised2019.07 | — | 78.3 | 91.1 | 2.3 | 77.2 | 84.7 | 4.9 | 22 | — | — | |
| PDBLearning Protocol=Unsupervised2019.07 | — | 77.2 | 90.1 | 0.9 | 74.5 | 84.4 | -0.2 | 29.1 | — | — | |
| SAGESupervision=false, RGB=true, Flow=true, Runtime=0.9s2021.04 | — | 42.6 | — | — | — | — | — | — | — | — | |
| SFLOptical flow=FlowNetS [20]2022.09 | — | — | — | — | — | — | — | — | 67.4 | 66.7 | |
| SFLSupervision=true, RGB=true, Flow=true, Resolution=854 x 480, Runtime=7.9s2021.04 | — | 67.4 | — | — | — | — | — | — | — | — | |
| SLT-Net2022.03 | — | 77.96 | 95.49 | — | 78.65 | 92.08 | — | — | — | — | |
| VITAE2022.09 | — | — | — | — | — | — | — | — | 89.2 | 90.4 |