Research map / 3D vision / Depth and stereo
SceneFlowFields++: Multi-frame Matching, Visibility Prediction, and Robust Interpolation for Scene Flow Estimation
René Schuster, Oliver Wasenmüller, Christian Unger, Georg Kuschk, Didier Stricker
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 |
|---|---|---|---|
| replace epic3d with robust ric3d interpolation (ablation) | dual+epic3d+kitti · KITTI scene flow training (200 images) | SF-all outliers (%): 26.12 -> 13.74 | Improved |
| add multi-frame matching with visibility reasoning (ablation) | dual+ric3d+kitti · KITTI scene flow training (200 images) | SF-all outliers (%): 13.74 -> 12.07 | Improved |
| use KITTI-trained edge detector instead of BSDS (ablation) | dual+epic3d+bsds · KITTI scene flow training (200 images) | SF-all outliers (%): 27.73 -> 26.12 | Improved |
| use universal mixed edge detector instead of KITTI-specific one (ablation) | multi+ric3d+kitti · KITTI scene flow training (200 images) | SF-all outliers (%): 12.07 -> 12.27 | No clear effect |
| add variational refinement (ablation) | multi+ric3d+kitti · KITTI scene flow training (200 images) | SF-all outliers (%): 12.07 -> 12.54 | Hurt |
| add ego-motion refinement after variational (ablation) | multi+ric3d+kitti+var · KITTI scene flow training (200 images) | SF-all outliers (%): 12.54 -> 12.62 | No clear effect |
| add ego-motion refinement to dual epic3d pipeline (ablation) | dual+epic3d+kitti+var · KITTI scene flow training (200 images) | SF-all outliers (%): 25.18 -> 14.53 | Improved |
| add variational refinement (ablation) | dual+ric3d+kitti · KITTI scene flow training (200 images) | SF-all outliers (%): 13.74 -> 13.77 | No clear effect |
| use multi-frame SFF++ instead of dual-frame SFF at the interpolation stage | SFF · KITTI training, interpolated (all pixels) | SF outliers (%): 26.1 -> 12.7 | Improved |
| use SFF++ instead of SFF on the KITTI benchmark | SFF · KITTI Scene Flow Benchmark (test) | SF-all outliers (%): 15.78 -> 14.59 | Improved |
| use SFF++ instead of SFF on Sintel disparity | SFF · MPI Sintel (average, disparity) | outliers EPE>3px (%): 18.15 -> 13.60 | Improved |
| use SFF++ instead of SFF on Sintel optical flow | SFF · MPI Sintel (average, optical flow) | outliers EPE>3px (%): 22.20 -> 18.47 | Improved |
Closest papers by meaning
- Flow Fields: Dense Correspondence Fields for Highly Accurate Large Displacement Optical Flow Estimation (TPAMI 2019 · similarity 0.91)
- Deep Scene Flow Learning: From 2D Images to 3D Point Clouds (TPAMI 2023 · similarity 0.90)
- Dense Flow from Adaptive Correspondence (NeurIPS 2026 · similarity 0.89)
- DifFlow3D: Toward Robust Uncertainty-Aware Scene Flow Estimation with Iterative Diffusion-Based Refinement (CVPR 2024 · similarity 0.89)
- Sparse Global Matching for Video Frame Interpolation with Large Motion (CVPR 2024 · similarity 0.89)
- DeltaFlow: An Efficient Multi-frame Scene Flow Estimation Method (NeurIPS 2025 · similarity 0.88)
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.
