Research map / Architectures and efficiency / CNN design
Multi-Level Context Ultra-Aggregation for Stereo Matching
Check 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 |
|---|---|---|---|
| apply MCUA multi-level context aggregation to PSM-Net | PSM-Net · Scene Flow test | EPE: 1.09 -> 0.56 | Helped |
| apply MCUA multi-level context aggregation to PSM-Net | PSM-Net · KITTI2015 test | D1-all (All, %): 2.32 -> 2.14 | Helped |
| apply MCUA multi-level context aggregation to PSM-Net | PSM-Net · KITTI2012 test | >3px All (%): 1.89 -> 1.70 | Helped |
| add the residual refinement module to MCUA (ablation) | MCUA · KITTI2015 test | D1-all (All, %): 2.14 -> 2.09 | Helped |
| add the residual refinement module to MCUA (ablation) | MCUA · KITTI2012 test | >3px All (%): 1.70 -> 1.64 | Helped |
| add the residual refinement module to MCUA (ablation) | MCUA · KITTI2015 test foreground | D1-fg (%): 4.38 -> 4.27 | Helped |
| use dense intra-level aggregation instead of plain PSM-Net matching-cost features (ablation) | PSM-Net · Scene Flow validation | EPE: 1.119 -> 0.794 | Helped |
| use DLA aggregation instead of plain PSM-Net matching-cost features (ablation) | PSM-Net · Scene Flow validation | EPE: 1.119 -> 0.806 | Helped |
| untie the child module parameters from the backbone (ablation) | MCUA · Scene Flow validation | EPE: 0.758 -> 0.755 | No clear effect |
| remove inter-level aggregation through the child module (ablation) | MCUA · Scene Flow validation | EPE: 0.758 -> 0.777 | Hurt |
| add inter-level aggregation through the child module (ablation) | intra-level-only stereo matching · Scene Flow validation | EPE: 0.761 -> 0.758 | Helped |
| add inter-level aggregation through the child module (ablation) | intra-level-only stereo matching · KITTI2015 validation | VE (%): 1.628 -> 1.579 | Helped |
Closest papers by meaning
- Deep Stereo Matching With Explicit Cost Aggregation Sub-Architecture (AAAI 2018 · similarity 0.91)
- Adaptive Unimodal Cost Volume Filtering for Deep Stereo Matching (AAAI 2020 · similarity 0.90)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.90)
- Hierarchical Object-Aware Dual-Level Contrastive Learning for Domain Generalized Stereo Matching (NeurIPS 2024 · similarity 0.88)
- IINet: Implicit Intra-inter Information Fusion for Real-Time Stereo Matching (AAAI 2024 · similarity 0.87)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.87)
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.