Research map / Label-efficient and robust learning / Recognition and long-tail
Learning for Disparity Estimation Through Feature Constancy
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 |
|---|---|---|---|
| use iResNet-i2 instead of CRL | stereo disparity estimation · KITTI 2015 test, all pixels | D1-all error (%): 2.67 -> 2.44 | Helped |
| use iResNet-i2 instead of GC-NET | stereo disparity estimation · KITTI 2012 test, Out-All > 3px | erroneous pixels (%): 2.30 -> 2.16 | Helped |
| use iResNet instead of CRL | Scene Flow disparity estimation · Scene Flow test | EPE: 1.60 -> 1.40 | Helped |
| add disparity refinement with feature constancy (ablation) | DES-net on Scene Flow · Scene Flow test | EPE: 2.81 -> 2.50 | Helped |
| add multi-scale skip connections instead of single-scale skip connections (ablation) | iResNet · Scene Flow test | EPE: 2.55 -> 2.50 | Helped |
| add feature correlation and reconstruction error to disparity refinement (ablation) | DES-net + DRS-net · Scene Flow test | EPE: 2.72 -> 2.50 | Helped |
| add reconstruction error to disparity refinement (ablation) | iResNet · Scene Flow test | EPE: 2.70 -> 2.50 | Helped |
| add feature correlation to disparity refinement (ablation) | iResNet · Scene Flow test | EPE: 2.56 -> 2.50 | Helped |
| add initial disparity as refinement input (ablation) | iResNet · Scene Flow test | EPE: 2.61 -> 2.50 | Helped |
| add a second feature-constancy refinement iteration (ablation) | iResNet · Scene Flow test | EPE: 2.50 -> 2.46 | Helped |
| add a third feature-constancy refinement iteration (ablation) | iResNet · Scene Flow test | EPE: 2.46 -> 2.45 | Helped |
Closest papers by meaning
- Local Similarity Pattern and Cost Self-Reassembling for Deep Stereo Matching Networks (AAAI 2022 · similarity 0.91)
- Efficient Deep Learning for Stereo Matching (CVPR 2016 · similarity 0.90)
- Deep Stereo Matching With Explicit Cost Aggregation Sub-Architecture (AAAI 2018 · similarity 0.90)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.90)
- Improved Stereo Matching With Constant Highway Networks and Reflective Confidence Learning (CVPR 2017 · similarity 0.89)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.89)
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.