Research map / Architectures and efficiency / CNN design
GA-Net: Guided Aggregation Net for End-To-End 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 |
|---|---|---|---|
| Use GA-Net-15 instead of PSMNet | Stereo matching on Scene Flow · Scene Flow test | EPE: 1.09 -> 0.84 | Helped |
| Use GA-Net-15 instead of PSMNet | Stereo matching on KITTI 2012 · KITTI 2012 test | Avg-All endpoint error: 0.6 px -> 0.5 px | Helped |
| Use GA-Net-15 instead of PSMNet | Stereo matching on KITTI 2015 · KITTI 2015 test | All-areas Avg All error rate: 2.32% -> 2.03% | Helped |
| Add a stacked feature-extraction block (ablation) | GA-Net · Scene Flow test | EPE: 1.26 -> 1.19 | Helped |
| Add dense concatenations to the stacked feature block (ablation) | GA-Net · Scene Flow test | EPE: 1.19 -> 1.14 | Helped |
| Add one SGA layer (ablation) | GA-Net with densely concatenated features · Scene Flow test | EPE: 1.14 -> 1.05 | Helped |
| Increase SGA layers from three to four (ablation) | GA-Net with densely concatenated features · Scene Flow test | EPE: 0.90 -> 0.89 | Helped |
| Add LGA to a three-SGA configuration (ablation) | GA-Net · Scene Flow test | EPE: 0.90 -> 0.84 | Helped |
| Add LGA to a three-SGA configuration (ablation) | GA-Net · KITTI 2015 validation | 3-pixel error rate: 2.85 -> 2.71 | Helped |
| Use two GA layers and two 3D convolutions instead of GC-Net's 3D-convolution aggregation | GA-Net-2 · Scene Flow test | EPE: 1.80 -> 1.51 | Helped |
| Add guided aggregation layers (ablation) | GA-Net models compared with the same architectures without GA layers · Scene Flow | EPE | Helped |
Closest papers by meaning
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.93)
- Deep Stereo Matching With Explicit Cost Aggregation Sub-Architecture (AAAI 2018 · similarity 0.93)
- GraftNet: Towards Domain Generalized Stereo Matching With a Broad-Spectrum and Task-Oriented Feature (CVPR 2022 · similarity 0.89)
- Guided Stereo Matching (CVPR 2019 · similarity 0.89)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.87)
- Hierarchical Object-Aware Dual-Level Contrastive Learning for Domain Generalized Stereo Matching (NeurIPS 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.