Research map / 3D vision / Depth and stereo
Group-Wise Correlation Stereo Network
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 |
|---|---|---|---|
| Replace concatenation-only matching costs with 40-group correlation volume | GwcNet on KITTI 2015 · KITTI 2015 test | D1-all (%): 2.32 -> 2.11 | Helped |
| Add group-wise correlation to the concatenation cost volume | GwcNet on KITTI 2012 · KITTI 2012 test | >3px All (%): 1.89 -> 1.70 | Helped |
| Use 40-group correlation instead of concatenation-only volume (ablation) | Base model on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.308 -> 1.212 | Helped |
| Use one correlation group (full correlation) rather than 40 groups (ablation) | Base model on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.212 -> 1.369 | Hurt |
| Increase group count from 10 to 20 (ablation) | Gwc-Base on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.230 -> 1.216 | Helped |
| Increase group count from 20 to 40 (ablation) | Gwc-Base on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.216 -> 1.212 | Helped |
| Increase group count from 40 to 80 (ablation) | Gwc-Base on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.212 -> 1.214 | No clear effect |
| Increase group count from 80 to 160 (ablation) | Gwc-Base on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.214 -> 1.188 | Helped |
| Add concatenation volume to 40-group correlation (ablation) | Base model on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 1.212 -> 1.127 | Helped |
| Replace PSMNet's hourglass with the proposed hourglass modifications (ablation) | Concatenation-volume stereo model on Scene Flow Finalpass · Scene Flow Finalpass | EPE (px): 0.876 -> 0.808 | Helped |
| Use the proposed hourglass modifications instead of the original hourglass (ablation) | Concatenation-volume stereo model · Scene Flow Finalpass | Time (ms): 241.0 -> 198.3 | Helped |
| Add group-wise correlation to concatenation volume (ablation) | GwcNet on KITTI 2012 validation · KITTI 2012 validation | EPE (px): 0.691 -> 0.659 | Helped |
Closest papers by meaning
- Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume (AAAI 2020 · similarity 0.90)
- Practical Stereo Matching via Cascaded Recurrent Network With Adaptive Correlation (CVPR 2022 · similarity 0.89)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.88)
- Learning for Disparity Estimation Through Feature Constancy (CVPR 2018 · similarity 0.88)
- Deep Stereo Matching With Explicit Cost Aggregation Sub-Architecture (AAAI 2018 · similarity 0.87)
- Local Similarity Pattern and Cost Self-Reassembling for Deep Stereo Matching Networks (AAAI 2022 · 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.