Research map / Probabilistic and scientific modeling / Molecules and proteins
Sensor-Guided Optical Flow
Matteo Poggi, Filippo Aleotti, Stefano Mattoccia
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 |
|---|---|---|---|
| run correlation at quarter resolution instead of eighth (QRAFT) (ablation) | RAFT · KITTI 142 (trained C+T, no guide) | EPE: 6.34 -> 6.21 | Improved |
| train RAFT with 3x larger batch | RAFT · KITTI 142 (trained C+T) | EPE: 6.34 -> 4.14 | Improved |
| add simulated sparse flow guide | QRAFT · KITTI 142 (trained C+T) | EPE: 6.21 -> 3.15 | Improved |
| add simulated sparse flow guide | RAFT (x3 batch) · KITTI 142 (trained C+T) | EPE: 4.14 -> 4.53 | Hurt |
| add simulated sparse flow guide | RAFT (x3 batch) · KITTI 142 (trained C) | EPE: 8.77 -> 5.39 | Improved |
| add simulated sparse flow guide | QRAFT · KITTI 142 (trained C+T+K) | EPE: 2.58 -> 1.22 | Improved |
| add simulated sparse flow guide | QRAFT · Sintel Clean (trained C+T+K) | EPE: 5.03 -> 1.63 | Improved |
| filter EgoFlow guide with forward-backward consistency (ablation) | EgoFlow guide · KITTI 142 (guide accuracy) | guide EPE: 3.25 -> 2.39 | Improved |
| add RICFlow with MaskRCNN-based static/dynamic selection to filtered EgoFlow (ablation) | flow guide · KITTI 142 (guide accuracy) | guide EPE: 2.39 -> 0.80 | Improved |
| use MaskRCNN semantics instead of motion mask to pick EgoFlow vs RICFlow (ablation) | flow guide · KITTI 142 (guide accuracy) | guide EPE: 1.32 -> 0.80 | Improved |
| use LIDAR-derived flow guide | QRAFT · KITTI 142 (trained C) | EPE: 9.61 -> 5.88 | Improved |
| use LIDAR-derived flow guide | QRAFT · KITTI 142 (trained C+T+K) | EPE: 2.58 -> 2.08 | Improved |
Closest papers by meaning
- Uncertainty Estimates and Multi-Hypotheses Networks for Optical Flow (ECCV 2018 · similarity 0.87)
- Learning Optical Flow From Still Images (CVPR 2021 · similarity 0.86)
- Learning Optical Flow From a Few Matches (CVPR 2021 · similarity 0.86)
- Slow Flow: Exploiting High-Speed Cameras for Accurate and Diverse Optical Flow Reference Data (CVPR 2017 · similarity 0.86)
- Flow Field Reconstruction with Sensor Placement Policy Learning (NeurIPS 2025 · similarity 0.85)
- FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024 · similarity 0.84)
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.
