Research map / 3D vision / Depth and stereo
MPI-Flow: Learning Realistic Optical Flow with Multiplane Images
Yingping Liang, Jiaming Liu, Debing Zhang, Ying Fu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add independent object motion to MPI-Flow (ablation) | MPI-Flow · KITTI 15 (unlabeled) | EPE: 2.46 -> 1.88 | Improved |
| add independent object motion to MPI-Flow (ablation) | MPI-Flow · KITTI 12 | EPE: 1.20 -> 1.26 | Hurt |
| add depth-aware inpainting (ablation) | MPI-Flow · Sintel.F | EPE: 3.82 -> 3.71 | Improved |
| add depth-aware inpainting and multiple objects (ablation) | MPI-Flow · KITTI 15 (unlabeled) | EPE: 1.88 -> 1.92 | No clear effect |
| add multiple objects per image (ablation) | MPI-Flow · Sintel.C | EPE: 2.37 -> 2.20 | Improved |
| increase virtual camera motions per image from 1 to 40 (ablation) | MPI-Flow · KITTI 15 | EPE: 2.02 -> 1.80 | Improved |
| increase dynamic objects per image to 2 (ablation) | MPI-Flow · KITTI 15 | EPE: 1.91 -> 2.02 | Hurt |
| increase source images from 20K to 120K (ablation) | MF-COCO · KITTI 15 | EPE: 3.68 -> 3.41 | Improved |
| narrow t_x and t_y range to -0.1~0.1 (ablation) | MPI-Flow · KITTI 12 | EPE: 1.18 -> 1.32 | Hurt |
| use MPI-Flow instead of Depthstillation data | RAFT from scratch (COCO) · KITTI 15 | Fl: 12.44 -> 11.95 | Improved |
| use MPI-Flow data instead of Depthstillation data for PWC-Net | PWC-Net from scratch (DAVIS) · KITTI 15 | EPE: 6.88 -> 6.92 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports MPI-Flow as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Flow-Anything: Learning Real-World Optical Flow Estimation From Large-Scale Single-View Images (TPAMI 2025 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Flow-Anything: Learning Real-World Optical Flow Estimation From Large-Scale Single-View Images (TPAMI 2025 · similarity 0.91)
- Learning Optical Flow From Still Images (CVPR 2021 · similarity 0.90)
- RealFlow: EM-Based Realistic Optical Flow Dataset Generation from Videos (ECCV 2022 · similarity 0.90)
- Slow Flow: Exploiting High-Speed Cameras for Accurate and Diverse Optical Flow Reference Data (CVPR 2017 · similarity 0.89)
- MemFlow: Optical Flow Estimation and Prediction with Memory (CVPR 2024 · similarity 0.87)
- Flow-NeRF: Joint Learning of Geometry, Poses, and Dense Flow within Unified Neural Representations (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
