danini/graph-cut-ransac
C++ implementation of the Graph-Cut RANSAC algorithm for robust estimation of geometric models (homography, fundamental matrix, essential matrix) from image data.

This repository provides an implementation of the Graph-Cut RANSAC algorithm published at CVPR 2018. The algorithm performs robust estimation of geometric transformations between image pairs by iteratively sampling correspondences and optimizing via graph-cut. It estimates homographies, fundamental matrices, and essential matrices used in camera pose estimation and multi-view geometry tasks. A Python wrapper (pygcransac) is also available via PyPI.