Research map / Label-efficient and robust learning / Recognition and long-tail
Matching neural paths: transfer from recognition to correspondence search
Nikolay Savinov, Lubor Ladicky, Marc Pollefeys
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 |
|---|---|---|---|
| use hierarchical path matching instead of stacked-feature correlation | VGG-16 transfer for stereo correspondence · KITTI 2012 validation | Err_3 (%): 5.23 -> 3.94 | Improved |
| use hierarchical path matching instead of stacked-feature correlation | VGG-16 transfer for stereo correspondence · KITTI 2015 validation | Err_3 (%): 6.16 -> 4.78 | Improved |
| use hierarchical path matching instead of the best listed learning-based baseline, acrt | VGG-16 transfer for stereo correspondence · KITTI 2012 validation | Err_3 (%): 2.61 -> 3.94 | Hurt |
| use hierarchical path matching instead of the best listed learning-based baseline, acrt | VGG-16 transfer for stereo correspondence · KITTI 2015 validation | Err_3 (%): 3.25 -> 4.78 | Hurt |
| add the full feature hierarchy rather than using only layer 2 (ablation) | ours(2, 8) · KITTI 2012 validation | Err_3 (%): 5.28 -> 3.94 | Improved |
| add the full feature hierarchy rather than using only layers 2 through 3 (ablation) | ours(2, 8) · KITTI 2012 validation | Err_3 (%): 5.41 -> 3.94 | Improved |
| use full convolutional receptive-field coverage rather than only central arcs (ablation) | ours(2, 8) · KITTI 2012 validation | Err_3 (%): 4.02 -> 3.94 | Improved |
| use sum/product aggregation rather than the alternative max-based aggregation (ablation) | neural-path matching · stereo correspondence | Err_3 (%) | Improved |
Closest papers by meaning
- Neural Matching Fields: Implicit Representation of Matching Fields for Visual Correspondence (NeurIPS 2022 · similarity 0.89)
- Learning Semantic Correspondence Exploiting an Object-Level Prior (TPAMI 2020 · similarity 0.89)
- Semantic Correspondence: Unified Benchmarking and a Strong Baseline (TPAMI 2025 · similarity 0.88)
- COTR: Correspondence Transformer for Matching Across Images (ICCV 2021 · similarity 0.88)
- Hierarchical Object-Aware Dual-Level Contrastive Learning for Domain Generalized Stereo Matching (NeurIPS 2024 · similarity 0.87)
- MATCHA: Towards Matching Anything (CVPR 2025 · 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.
