Research map / 3D vision / Depth and stereo
On the confidence of stereo matching in a deep-learning era: a quantitative evaluation
Matteo Poggi, Seungryong Kim, Fabio Tosi, Sunok Kim, Filippo Aleotti, Dongbo Min, Kwanghoon Sohn, 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 |
|---|---|---|---|
| train learned confidence measure on real KITTI 2012 instead of synthetic Driving | LAF · KITTI 2012 | AUC (x100, lower better): 6.87 -> 5.33 | Improved |
| train on KITTI 2012 instead of Driving | MPN · KITTI 2012 | AUC (x100, lower better): 6.53 -> 5.58 | Improved |
| train on KITTI 2012 instead of Driving | MPN · ETH3D | AUC (x100, lower better): 7.08 -> 6.23 | Improved |
| train on KITTI 2012 instead of Driving | LAF · ETH3D | AUC (x100, lower better): 9.12 -> 9.50 | Hurt |
| train on KITTI 2012 instead of Driving | PBCP_d · Middlebury | AUC (x100, lower better): 14.24 -> 15.01 | Hurt |
| use learned measure LAF instead of best hand-crafted DA31 | DA31 · KITTI 2015 | AUC (x100, lower better): 6.45 -> 5.35 | Improved |
| use learned measure FA instead of hand-crafted DA31 | DA31 · Middlebury | AUC (x100, lower better): 12.92 -> 13.99 | Hurt |
| use learned measure LAF instead of DA31 | DA31 · Driving | AUC (x100, lower better): 23.12 -> 13.76 | Improved |
| use WTA selection instead of soft-argmax in GANet | GANet · KITTI 2012 (bad3) | bad error (%): 9.51% -> 8.47% | Improved |
| use WTA selection instead of soft-argmax in GANet | GANet · ETH3D (bad1) | bad error (%): 8.73% -> 10.80% | Hurt |
Closest papers by meaning
- Learning to Predict Stereo Reliability Enforcing Local Consistency of Confidence Maps (CVPR 2017 · similarity 0.93)
- Modeling Stereo-Confidence out of the End-to-End Stereo-Matching Network via Disparity Plane Sweep (AAAI 2024 · similarity 0.93)
- A Survey on Deep Learning Techniques for Stereo-Based Depth Estimation (TPAMI 2020 · similarity 0.93)
- A Survey on Deep Stereo Matching in the Twenties (IJCV 2025 · similarity 0.92)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.92)
- On the Synergies between Machine Learning and Binocular Stereo for Depth Estimation from Images: a Survey (TPAMI 2021 · similarity 0.92)
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.
