Video Panoptic Segmentation on KITTI-STEP (val)
74STQVideo K-Net
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Video K-NetBackbone=Swin-large, Optical Flow=false2022.04 | 74 | 73 | 75 | 56 | |
| Video K-NetBackbone=Swin-base, Optical Flow=false2022.04 | 73 | 72 | 73 | 53 | |
| Video K-NetBackbone=Swin-base, OF=false2023.03 | 73 | 72 | 73 | 53 | |
| TarViSShared Model=true, Backbone=Swin-L2023.01 | 72 | 72 | 73 | — | |
| Tube-LinkBackbone=Swin-base, OF=false2023.03 | 72 | 69 | 74 | 56 | |
| Video K-NetBackbone=ResNet50, Optical Flow=false2022.04 | 71 | 70 | 71 | 46 | |
| TarViSShared Model=true, Backbone=Swin-T2023.01 | 71 | 71 | 70 | — | |
| Video K-NetBackbone=ResNet50, OF=false2023.03 | 71 | 70 | 71 | 46 | |
| TubeFormer-DeepLabBackbone=ResNet-50 + axial, Optical Flow=false2022.04 | 70 | 64 | 76 | 51 | |
| TarViSShared Model=true, Backbone=R-502023.01 | 70 | 70 | 69 | — | |
| TubeFormer-DeepLabBackbone=ResNet-50 + Axial, OF=false2023.03 | 70 | 64 | 76 | 51 | |
| Tube-LinkBackbone=ResNet50, OF=false2023.03 | 68 | 67 | 69 | 51 | |
| Panoptic DeepLab + Mask PropagationBackbone=ResNet50, Optical Flow=true2022.04 | 67 | 63 | 71 | 44 | |
| Mask PropagationShared Model=false2023.01 | 67 | 63 | 71 | — | |
| P + Mask PropagationBackbone=ResNet50, OF=false2023.03 | 67 | 63 | 71 | 44 | |
| Panoptic DeepLab + SORTBackbone=ResNet50, Optical Flow=false2022.04 | 59 | 50 | 71 | 42 | |
| Panoptic DeepLab + IoU Assoc.Backbone=ResNet50, Optical Flow=false2022.04 | 58 | 47 | 71 | 44 | |
| Motion-DeeplabBackbone=ResNet50, Optical Flow=false2022.04 | 58 | 51 | 67 | 40 | |
| Motion-DeepLabBackbone=ResNet50, OF=true2023.03 | 58 | 51 | 67 | 40 | |
| VPSNetBackbone=ResNet50, Optical Flow=true2022.04 | 56 | 52 | 61 | 43 | |
| VPSNetShared Model=false2023.01 | 56 | 52 | 61 | — | |
| VPSNetBackbone=ResNet50, OF=true2023.03 | 56 | 52 | 61 | 43 |