Research map / 3D vision / Depth and stereo
Self-adapting confidence estimation for stereo
Matteo Poggi, Filippo Aleotti, Fabio Tosi, Giulio Zaccaroni, 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 |
|---|---|---|---|
| add disparity-agreement and uniqueness positive criteria to the reprojection cue (ablation) | OTB ConfNet · KITTI 2012 val (Census-CBCA) | AUC (lower better): 0.077 -> 0.070 | Improved |
| add negative criteria A^q and U^q to the label set (ablation) | OTB ConfNet · KITTI 2012 val (Census-CBCA) | AUC (lower better): 0.070 -> 0.071 | Hurt |
| add negative criteria A^q and U^q to the label set (ablation) | OTB ConfNet · KITTI 2012 val (Census-SGM) | AUC (lower better): 0.029 -> 0.028 | Improved |
| add all positive criteria with only T^q negative instead of all six, tested on unseen Middlebury (ablation) | OTB CCNN · Middlebury 2014 (Census-CBCA) | AUC (lower better): 0.153 -> 0.116 | Improved |
| train confidence network on single reprojection cue vs. the reprojection baseline used as supervision (ablation) | CCNN · KITTI 2012 val (Census-CBCA) | AUC (lower better): 0.210 -> 0.080 | Improved |
| replace WILD supervision with OTB supervision | ConfNet · KITTI 2012 val (Census-CBCA) | AUC (lower better): 0.089 -> 0.070 | Improved |
| replace SELF or WILD supervision with OTB on accurate SGM algorithms | CCNN · KITTI 2012 val (Census-SGM) | AUC (lower better): WILD 0.026, OTB 0.029 | No clear effect |
| replace SELF supervision with OTB supervision | CCNN · Middlebury 2014 (Census-CBCA) | AUC (lower better): 0.163 -> 0.116 | Improved |
| add online adaptation of the confidence network | OTB ConfNet · DrivingStereo sequence (Census-SGM) | AUC (lower better): 0.072 -> 0.064 | Improved |
| add online adaptation of the confidence network | OTB ConfNet · DrivingStereo sequence (MADNet) | AUC (lower better): 0.139 -> 0.126 | Improved |
| add online adaptation of the confidence network | OTB ConfNet · DrivingStereo sequence (GANet) | AUC (lower better): 0.019 -> 0.015 | Improved |
| use OTB self-supervision instead of full ground-truth supervision, evaluated on transfer to ETH3D | CCNN · ETH3D (MCCNN-fst-SGM) | AUC (lower better): 0.046 -> 0.042 | Improved |
Closest papers by meaning
- Using Self-Contradiction to Learn Confidence Measures in Stereo Vision (CVPR 2016 · similarity 0.92)
- Learning to Predict Stereo Reliability Enforcing Local Consistency of Confidence Maps (CVPR 2017 · similarity 0.92)
- Stereo Confidence Estimation via Locally Adaptive Fusion and Knowledge Distillation (TPAMI 2022 · similarity 0.91)
- Modeling Stereo-Confidence out of the End-to-End Stereo-Matching Network via Disparity Plane Sweep (AAAI 2024 · similarity 0.91)
- Quantitative Evaluation of Confidence Measures in a Machine Learning World (ICCV 2017 · similarity 0.91)
- On the confidence of stereo matching in a deep-learning era: a quantitative evaluation (TPAMI 2021 · similarity 0.91)
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.
