Research map / Label-efficient and robust learning / Self-supervised and contrastive
TeFlow: Enabling Multi-frame Supervision for Self-Supervised Feed-forward Scene Flow Estimation
Qingwen Zhang, Chenhan Jiang, Xiaomeng Zhu, Yunqi Miao, Yushan Zhang, Olov Andersson, Patric Jensfelt
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 |
|---|---|---|---|
| use 5 input frames instead of 2 (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.353 -> 0.265 | Improved |
| use 5 input frames instead of 8 (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.300 -> 0.265 | Improved |
| add static loss to geometric loss (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.386 -> 0.458 | Hurt |
| add dynamic cluster loss to geometric and static losses (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.458 -> 0.265 | Improved |
| add cluster-level term to point-level loss (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.351 -> 0.265 | Improved |
| add point-level term to cluster-level loss (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.356 -> 0.265 | Improved |
| add external geometric candidates to internal predictions (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.455 -> 0.265 | Improved |
| add internal candidate to external candidates (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.321 -> 0.265 | Improved |
| add directional consistency agreement matrix M (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.349 -> 0.265 | Improved |
| add reliability weights w (ablation) | TeFlow · Argoverse 2 val | three-way EPE mean: 5.70 -> 4.43 | Improved |
| add flow aggregation of consistent candidates (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.289 -> 0.265 | Improved |
| use DeltaFlow backbone instead of Flow4D (ablation) | TeFlow · Argoverse 2 val | dynamic bucket-normalized EPE: 0.330 -> 0.265 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VoteFlow — VoteFlow: Enforcing Local Rigidity in Self-Supervised Scene Flow (CVPR 2025) · better on 4 of 4 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
- DeltaFlow: An Efficient Multi-frame Scene Flow Estimation Method (NeurIPS 2025 · similarity 0.90)
- MESSENGER: Memory-Enhanced Sequential Scene Flow Estimation via Autoregressive Next-Frame Forecasting (NeurIPS 2026 · similarity 0.89)
- Self-Supervised Multi-Frame Monocular Scene Flow (CVPR 2021 · similarity 0.89)
- ZeroFlow: Scalable Scene Flow via Distillation (ICLR 2024 · similarity 0.89)
- SCOOP: Self-Supervised Correspondence and Optimization-Based Scene Flow (CVPR 2023 · similarity 0.88)
- StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences (NeurIPS 2024 · 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.
