Research map / Architectures and efficiency / RNNs and normalization
AutoFlow: Learning a Better Training Set for Optical Flow
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Pre-train on AutoFlow instead of FlyingChairs | PWC-Net · Sintel.clean | AEPE: 3.27 -> 2.17 | Helped |
| Pre-train on AutoFlow instead of FlyingChairs | PWC-Net · Sintel.final | AEPE: 4.42 -> 2.91 | Helped |
| Pre-train on AutoFlow instead of FlyingChairs | RAFT · KITTI | AEPE: 7.63 -> 4.23 | Helped |
| Pre-train with AutoFlow instead of Chairs→Things | RAFT · Sintel.clean | AEPE: 1.68 -> 1.95 | Hurt |
| Use AutoFlow pre-training and TF-RAFT fine-tuning schedule | RAFT · KITTI | Fl-all: 5.56% -> 4.78% | Helped |
| Add fog to AutoFlow rendering (ablation) | RAFT pre-trained on AutoFlow · Sintel.final | AEPE: 3.11 -> 2.75 | Helped |
| Add fog to AutoFlow rendering (ablation) | RAFT pre-trained on AutoFlow · Sintel.clean | AEPE: 2.07 -> 2.08 | No clear effect |
| Add motion blur to AutoFlow rendering (ablation) | RAFT pre-trained on AutoFlow · Sintel.final | AEPE: 3.77 -> 2.75 | Helped |
| Replace Davis images with OpenImages appearance images (ablation) | RAFT pre-trained on AutoFlow · Sintel.final | AEPE: 2.75 -> 2.85 | Hurt |
| Use 1080p instead of 720p appearance images (ablation) | RAFT pre-trained on AutoFlow · KITTI | AEPE: 4.66 -> 5.09 | Hurt |
| Use smooth rather than sharp polygon edges (ablation) | RAFT pre-trained on AutoFlow · Sintel.final | AEPE: 2.80 -> 2.75 | Helped |
| Add foreground object layers (ablation) | RAFT pre-trained on AutoFlow · Sintel.final | AEPE: 5.74 -> 3.07 | Helped |
Closest papers by meaning
- Taming generative video models for zero-shot optical flow extraction (NeurIPS 2025 · similarity 0.87)
- Efficient Dataset Distillation for Pre-Trained Self-Supervised Models via Statistical Flow Matching (NeurIPS 2026 · similarity 0.87)
- STARFlow: Scaling Latent Normalizing Flows for High-resolution Image Synthesis (NeurIPS 2025 · similarity 0.87)
- Mean Flows for One-step Generative Modeling (NeurIPS 2025 · similarity 0.87)
- ADFactory: An Effective Framework for Generalizing Optical Flow with NeRF (CVPR 2024 · similarity 0.86)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.86)
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.