Research map / Learning theory and algorithms / Prediction and decision losses
Re-weighting and 1-Point RANSAC-Based PnP Solution to Handle Outliers
Haoyin Zhou, Tao Zhang, Jagadeesan Jayender
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 R1PPnP soft re-weighting plus 1-point RANSAC instead of explicit outlier rejection (REPPnP) at high outlier rates | REPPnP · Synthetic ordinary 3D with outliers | robustness to outlier percentage: fails beyond 50% -> handles large percentages | Improved |
| use R1PPnP instead of REPPnP on quasi-singular cases | REPPnP · Synthetic quasi-singular with outliers | robustness to outlier percentage: fails at 5% -> handles large percentages | Improved |
| use R1PPnP instead of RANSAC+P3P/P4P with refinement | RANSAC+P3P + OPnP/ASPnP/RPnP · Synthetic with outliers | rotation and translation error | Improved |
| use R1PPnP instead of RANSAC+P3P/P4P at high outlier percentage | RANSAC+P3P/P4P · Synthetic with outliers | runtime | Improved |
| use R1PPnP instead of pure RANSAC+P3P when outliers below 30% | RANSAC+P3P · DTU real images | runtime | Hurt |
| use R1PPnP instead of OPnP as refinement | OPnP · DTU real images | accuracy and runtime | Mixed |
| use R1PPnP core algorithm instead of other PnP methods without outliers, rotation | OPnP, SDP · Synthetic outlier-free | rotation accuracy | No clear effect |
| use R1PPnP core on quasi-singular outlier-free translation | other PnP methods · Synthetic quasi-singular outlier-free | translation accuracy | Hurt |
| terminate on stagnating inlier count in a 20-iteration window instead of pose change between iterations (ablation) | R1PPnP · Synthetic 50% outliers | stopping speed and robustness | Improved |
| constrain det(R)=1 from the start instead of allowing mirror form and flipping lambda (ablation) | R1PPnP · Core algorithm | local minima | Improved |
| select the control point near the center of the point cloud (ablation) | R1PPnP · Core algorithm | convergence speed | Improved |
Closest papers by meaning
- Robustness Can Be Cheap: A Highly Efficient Approach to Discover Outliers under High Outlier Ratios (AAAI 2019 · similarity 0.85)
- Automatic Outlier Rectification via Optimal Transport (NeurIPS 2024 · similarity 0.84)
- ALTBI: Constructing Improved Outlier Detection Models via Optimization of Inlier-Memorization Effect (AAAI 2025 · similarity 0.83)
- RANSAC Scoring Done Right (NeurIPS 2026 · similarity 0.83)
- Outlier Synthesis via Hamiltonian Monte Carlo for Out-of-Distribution Detection (ICLR 2025 · similarity 0.83)
- DOS: Diverse Outlier Sampling for Out-of-Distribution Detection (ICLR 2024 · similarity 0.83)
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.
