Research map / Video / Video segmentation and tracking
VideoFlow: Exploiting Temporal Cues for Multi-frame Optical Flow Estimation
Xiaoyu Shi, Zhaoyang Huang, Weikang Bian, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, Hongsheng Li
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 |
|---|---|---|---|
| estimate bi-directional flows from the center frame instead of uni-directional warm-start (ablation) | TROF · KITTI-15 train | Fl-epe: 4.82 -> 4.60 | Improved |
| bi-directional estimation on KITTI-15 Fl-all (ablation) | TROF · KITTI-15 train | Fl-all (%): 17.48 -> 18.05 | Hurt |
| add recurrent fusion of bi-directional motion features (ablation) | TROF · KITTI-15 train | Fl-all (%): 18.05 -> 16.52 | Improved |
| add motion state feature M to MOP (ablation) | VideoFlow · FlyingThings val clean | AEPE: 1.61 -> 1.48 | Improved |
| add temporal GRU to pass motion state across all units (ablation) | VideoFlow MOP · KITTI-15 train | Fl-epe: 3.89 -> 4.04 | Hurt |
| increase from 3 frames to 5 frames | VideoFlow · Sintel test (final) | AEPE: 1.71 -> 1.65 | Improved |
| increase from 3 frames to 5 frames | VideoFlow · KITTI-15 test | Fl-all (%): 4.44 -> 3.65 | Improved |
| increase from 3 frames to 5 frames on Sintel train generalization | VideoFlow · Sintel train (final), FlyingThings only | AEPE: 2.19 -> 2.56 | Hurt |
| use multi-frame VideoFlow instead of warm-start three-frame methods | SKFlow* · Sintel test (clean) | AEPE: 1.28 -> 1.00 | Improved |
| use VideoFlow instead of FlowFormer++ on matched pixels, clean pass | FlowFormer++ · Sintel test (clean) matched | AEPE: 0.39 -> 0.40 | No clear effect |
Closest papers by meaning
- StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences (NeurIPS 2024 · similarity 0.93)
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.90)
- MemFlow: Optical Flow Estimation and Prediction with Memory (CVPR 2024 · similarity 0.89)
- FlowVid: Taming Imperfect Optical Flows for Consistent Video-to-Video Synthesis (CVPR 2024 · similarity 0.89)
- FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024 · similarity 0.88)
- 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.
