Optical Flow Estimation on Sintel Final unsupervised
Average End-Point Error 낮을수록 좋음
- 2020-03-29 — ARFlow-MV: Average End-Point Error 5.67
- 2020-12-01 — UpFlow: Average End-Point Error 5.32
| Rank | Model | Average End-Point Error | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | UpFlow | 5.32 | UPFlow: Upsampling Pyramid for Unsupervised Optical Flow Learning | twhui/LiteFlowNet3 · coolbeam/UPFlow_pytorch | 2020 |
| 2 | MDFlow | 5.46 | MDFlow: Unsupervised Optical Flow Learning by Reliable Mutual Knowledge Distillation | ltkong218/mdflow | 2022 |
| 3 | ARFlow-MV | 5.67 | Learning by Analogy: Reliable Supervision from Transformations for Unsupervised Optical Flow Estimation | lliuz/ARFlow · duke-vision/optical-flow-active-learning-release | 2020 |
| 4 | MDFlow-Fast | 5.99 | MDFlow: Unsupervised Optical Flow Learning by Reliable Mutual Knowledge Distillation | ltkong218/mdflow | 2022 |
| 5 | UFlow | 6.50 | What Matters in Unsupervised Optical Flow | google-research/google-research · junbongjang/contour-tracking · 2024-MindSpore-1/Code3 · +2 | 2020 |