Research map / Point clouds / Registration and matching
Discriminative Optimization: Theory and Applications to Computer Vision
Jayakorn Vongkulbhisal, Fernando N. de la Torre, João Paulo Costeira
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 DO learned on P4 data instead of fmincon with the matching nonconvex cost P4 | DO (SUM) · 1D unknown cost, test P4 | MAE: .0457 -> .0325 | Improved |
| use DO learned on P6 data instead of fmincon with the matching cost P6 | DO (SUM) · 1D unknown cost, test P6 | MAE: .1161 -> .0698 | Improved |
| use DO instead of fmincon with the correct convex cost P1 | DO (SUM) · 1D unknown cost, test P1 | MAE: .0000 -> .0137 | Hurt |
| use DO instead of fmincon with a mismatched cost on P3 | DO (SUM) · 1D unknown cost, test P3 | MAE: .1535 -> .0086 | Improved |
| train DO on the noise type that matches test noise (not mismatched) (ablation) | DO-SP / DO-RV · Image denoising impulse noise | PSNR | Improved |
| train DO on mixed SP and RV noise (DO-SPRV) (ablation) | DO · Image denoising impulse noise | PSNR | Mixed |
| add DO inlier selection before P3P+RANSAC | P3P+RANSAC · Camera pose estimation, real image | computation time: 8 times | Improved |
| add DO inlier selection before P3P+RANSAC at high outlier ratio | P3P+RANSAC · Camera pose estimation, synthetic | rotation error | Improved |
| use DO alone for pose without geometric constraints | DO · Camera pose estimation, synthetic | rotation error | Hurt |
| replace ICP with DO for object tracking in 3D point cloud | DO · 3D object tracking (kettle, hat) | tracking robustness: 40ms per frame | Improved |
| replace ICP with DO for registration on UWA range scans | DO · UWA range-scan registration | success rate | Improved |
Closest papers by meaning
- Discriminative Optimization: Theory and Applications to Point Cloud Registration (CVPR 2017 · similarity 0.97)
- Certifiably Optimal Outlier-Robust Geometric Perception: Semidefinite Relaxations and Scalable Global Optimization (TPAMI 2022 · similarity 0.87)
- Inverse Composition Discriminative Optimization for Point Cloud Registration (CVPR 2018 · similarity 0.86)
- Reinforced Feature Points: Optimizing Feature Detection and Description for a High-Level Task (CVPR 2020 · similarity 0.86)
- Learning To Solve Hard Minimal Problems (CVPR 2022 · similarity 0.85)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 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.
