Research map / 3D vision / Depth and stereo
Iterative Geometry Encoding Volume for Stereo Matching
Read the paperCheck 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 |
|---|---|---|---|
| Replace all-pairs correlations with GEV (ablation) | RAFT-Stereo with MobileNetV2 100 · Scene Flow test | EPE (px): 0.56 -> 0.51 | Helped |
| Add initial disparity regressed from GEV (ablation) | GEV model · Scene Flow test | EPE (px): 0.51 -> 0.50 | Helped |
| Add GEV supervision (ablation) | GEV model with initial disparity · Scene Flow test | EPE (px): 0.50 -> 0.48 | Helped |
| Add all-pairs correlations to GEV (ablation) | GEV model with supervised initialization · Scene Flow test | EPE (px): 0.48 -> 0.47 | Helped |
| Use 1/8-resolution GEV instead of 1/4 (ablation) | IGEV-Stereo · Scene Flow test | EPE (px): 0.47 -> 0.49 | Hurt |
| Replace MobileNetV2 100 with MobileNetV2 120d (ablation) | IGEV-Stereo · Scene Flow test | EPE (px): 0.47 -> 0.46 | Helped |
| Replace MobileNetV2 100 with ConvNeXt-B (ablation) | IGEV-Stereo · Scene Flow test | EPE (px): 0.47 -> 0.45 | Helped |
| Use the full geometry volume approach rather than RAFT-Stereo | Stereo matching · Scene Flow test | EPE (px), 3 iterations: 0.95 -> 0.58 | Helped |
| Increase full-model update iterations from 1 to 32 (ablation) | IGEV-Stereo · Scene Flow test | EPE (px): 0.66 -> 0.47 | Helped |
| Increase RAFT-Stereo update iterations from 8 to 32 (ablation) | RAFT-Stereo · Scene Flow test | EPE (px): 0.66 -> 0.61 | Helped |
| Train on Scene Flow and evaluate zero-shot on ETH3D rather than RAFT-Stereo | Stereo matching · ETH3D | 1-pixel error rate: 3.2 -> 3.6 | Hurt |
Closest papers by meaning
- Generalized Geometry Encoding Volume for Real-time Stereo Matching (AAAI 2026 · similarity 0.90)
- Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching (CVPR 2020 · similarity 0.88)
- IINet: Implicit Intra-inter Information Fusion for Real-Time Stereo Matching (AAAI 2024 · similarity 0.88)
- IterMVS: Iterative Probability Estimation for Efficient Multi-View Stereo (CVPR 2022 · similarity 0.88)
- Any-Stereo: Arbitrary Scale Disparity Estimation for Iterative Stereo Matching (AAAI 2024 · similarity 0.87)
- All-in-One: Transferring Vision Foundation Models into Stereo Matching (AAAI 2025 · 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.