Research map / 3D vision / Depth and stereo
iMatching: Imperative Correspondence Learning
Zitong Zhan, Dasong Gao, Yun-Jou Lin, Youjie Xia, Chen Wang
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 |
|---|---|---|---|
| finetune CAPS with iMatching self-supervision | CAPS · TartanAir feature matching | MMA @1px: 26.9 -> 49.0 | Improved |
| finetune ASpanFormer with iMatching | ASpanFormer · TartanAir feature matching | MMA @1px: 51.5 -> 58.4 | Improved |
| finetune poorly initialized Patch2Pix with iMatching | Patch2Pix · TartanAir feature matching | MMA @1px: 0.8 -> 7.1 | Improved |
| finetune DKM with iMatching | DKM · ETH3D-SLAM relative pose | AUC@5deg: 58.2 -> 76.7 | Improved |
| train CAPS with SGP instead of iMatching | CAPS · ETH3D-SLAM relative pose | AUC@5deg: 71.5 -> 70.9 | No clear effect |
| finetune CAPS with iMatching | CAPS · ETH3D-SLAM relative pose | AUC@5deg: 71.5 -> 75.3 | Improved |
| finetune DKM with iMatching on KITTI360 | DKM · KITTI360 pose estimation | AUC@5deg: 91.8 -> 92.6 | No clear effect |
| finetune ASpanFormer with iMatching on KITTI360 | ASpanFormer · KITTI360 pose estimation | AUC@5deg: 74.4 -> 80.6 | Improved |
| add outlier rejection in bundle adjustment (ablation) | iCAPS · TartanAir Oldtown Easy | MMA @1px: 42.6 -> 59.2 | Improved |
| run more LM iterations in bundle adjustment (3 steps to full) (ablation) | iCAPS · TartanAir SeasonForest Easy | MMA @1px: 22.3 -> 29.4 | Improved |
| use own differentiable BA instead of COLMAP reconstruction for supervision (ablation) | iCAPS · TartanAir Amusement Easy | MMA @1px: 31.4 -> 59.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SuperGlue — SuperGlue: Learning Feature Matching With Graph Neural Networks (CVPR 2020) · better on 2 of 2 shared results
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
- Self-Supervised Cross-View Correspondence with Predictive Cycle Consistency (CVPR 2025 · similarity 0.86)
- Scalable and Generalizable Correspondence Pruning via Geometry-Consistent Pre-Training (TPAMI 2026 · similarity 0.86)
- Joint-task Self-supervised Learning for Temporal Correspondence (NeurIPS 2019 · similarity 0.86)
- Learning SO(3)-Invariant Semantic Correspondence via Local Shape Transform (CVPR 2024 · similarity 0.84)
- From Variance to Veracity: Unbundling and Mitigating Gradient Variance in Differentiable Bundle Adjustment Layers (CVPR 2024 · similarity 0.84)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.84)
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.
