Research map / Architectures and efficiency / Attention and transformers
Attention Concatenation Volume for Accurate and Efficient 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 |
|---|---|---|---|
| add ordinary fixed-size patch matching (ablation) | GwcNet · Scene Flow | EPE (px): 0.76 -> 0.72 | Helped |
| replace ordinary patch matching with multi-level adaptive patch matching (ablation) | GwcNet · Scene Flow | EPE (px): 0.72 -> 0.64 | Helped |
| use patch matching to generate attention weights and filter the concatenation volume (ablation) | GwcNet · Scene Flow | EPE (px): 0.64 -> 0.57 | Helped |
| add a 3D hourglass to regularize the attention matching volume (ablation) | Gwc-mp-att · Scene Flow | EPE (px): 0.57 -> 0.52 | Helped |
| supervise attention disparity using ground truth (ablation) | Gwc-mp-att-hg · Scene Flow | EPE (px): 0.52 -> 0.46 | Helped |
| replace the combined volume with ACV (ablation) | GwcNet · Scene Flow | D1 (%): 2.71 -> 1.55 | Helped |
| reduce aggregation hourglasses from 3 to 2 (ablation) | Gwc-acv-3 · Scene Flow | D1 (%): 1.55 -> 1.59 | Hurt |
| reduce aggregation hourglasses from 3 to 1 (ablation) | Gwc-acv-3 · Scene Flow | D1 (%): 1.55 -> 1.79 | Hurt |
| remove all aggregation hourglass networks (ablation) | Gwc-acv-3 · Scene Flow | D1 (%): 1.55 -> 2.08 | Hurt |
| apply ACV to PSMNet | PSMNet · Scene Flow | EPE (px): 1.09 -> 0.63 | Helped |
| apply ACV to GwcNet | GwcNet · Scene Flow | EPE (px): 0.76 -> 0.46 | Helped |
| apply ACV to CFNet | CFNet · Scene Flow | EPE (px): 0.97 -> 0.83 | Helped |
Closest papers by meaning
- CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching (CVPR 2021 · similarity 0.89)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.88)
- Multi-Level Context Ultra-Aggregation for Stereo Matching (CVPR 2019 · similarity 0.88)
- Selective-Stereo: Adaptive Frequency Information Selection for Stereo Matching (CVPR 2024 · similarity 0.87)
- MoCha-Stereo: Motif Channel Attention Network for Stereo Matching (CVPR 2024 · similarity 0.86)
- IINet: Implicit Intra-inter Information Fusion for Real-Time Stereo Matching (AAAI 2024 · 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.