Research map / Architectures and efficiency / RNNs and normalization
Accurate and Efficient Stereo Matching via Attention Concatenation Volume
Gangwei Xu, Yun Wang, Junda Cheng, Jinhui Tang, Xin Yan Yang
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 attention weights to the concatenation volume (ablation) | GwcNet · Scene Flow | EPE (px): 0.76 -> 0.57 | Improved |
| add hourglass for attention weights generation (ablation) | Gwc-att · Scene Flow | EPE (px): 0.57 -> 0.52 | Improved |
| add supervision for attention weights (ablation) | Gwc-att-hg · Scene Flow | EPE (px): 0.52 -> 0.46 | Improved |
| add VAP (variance-aware prediction) to Fast-ACVNet (ablation) | Fast-ACVNet · Scene Flow | D1 (%): 2.51 -> 2.32 | Improved |
| reduce F2I sampled candidates from 48 to 16 (ablation) | Fast-ACVNet with VAP · Scene Flow | runtime (ms): 50 -> 35 | Mixed |
| use lightweight backbone and aggregation network (ablation) | ACVNet · Scene Flow | runtime (ms): 200 -> 62 | Mixed |
| add attention filter in Fast-ACV with one hourglass (ablation) | Fast-ACV · Scene Flow | EPE (px): 0.83 -> 0.64 | Improved |
| reduce hourglass number from 3 to 0 in ACV (ablation) | Gwc-acv · Scene Flow | EPE (px): 0.46 -> 0.59 | Hurt |
| replace GwcNet cost volume with ACV (PSMNet) | PSMNet · Scene Flow | EPE (px): 0.88 -> 0.63 | Improved |
| replace GwcNet cost volume with ACV | GwcNet · Scene Flow | EPE (px): 0.76 -> 0.46 | Improved |
| use Fast-ACV instead of ACV on GwcNet | GwcNet · Scene Flow | runtime (ms): 200 -> 81 | Mixed |
| integrate VAP into PSMNet (ablation) | PSMNet · Scene Flow | EPE (px): 0.88 -> 0.71 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- HITNet — HitNet: Hybrid Ternary Recurrent Neural Network (NeurIPS 2018) · better on 2 of 3 shared results
- AANet — AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020) · better on 2 of 2 shared results
- GANet — GA-Net: Guided Aggregation Net for End-To-End Stereo Matching (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Attention Concatenation Volume for Accurate and Efficient Stereo Matching (CVPR 2022 · similarity 0.96)
- CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching (CVPR 2021 · similarity 0.91)
- Stereo Matching Using Multi-Level Cost Volume and Multi-Scale Feature Constancy (TPAMI 2019 · similarity 0.91)
- Any-Stereo: Arbitrary Scale Disparity Estimation for Iterative Stereo Matching (AAAI 2024 · similarity 0.88)
- Selective-Stereo: Adaptive Frequency Information Selection for Stereo Matching (CVPR 2024 · similarity 0.88)
- IINet: Implicit Intra-inter Information Fusion for Real-Time Stereo Matching (AAAI 2024 · similarity 0.88)
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.
