New RANSAC scoring method sorts residuals before separating inliers and outliers

ducha_aiki · x · 2026-07-21

A paper by James Pritts, Felix Seegräber, and Kevin Köser argues that RANSAC scoring should be done by sorting residuals first and then scoring each inlier/outlier against the right statistical model.

Their tl;dr:

The attached figure shows the algorithm and compares methods across several geometric datasets, framing the contribution as a cleaner way to compute a scale-free marginal score.

Related event: New RANSAC Scoring Method Proposes Sorting Residuals First(2 posts)→

Original post →

More from Research

Research channel →