ducha-aiki/pydegensac

Advanced RANSAC (DEGENSAC) with bells and whistles for H and F estimation

What it solves

It provides a robust way to estimate homography and fundamental matrices from sparse correspondences between images, specifically addressing the problem of outliers and degenerate configurations that often plague standard RANSAC implementations.

How it works

The project is a Python wrapper for C++ implementations of LO-RANSAC (Locally Optimized RANSAC) and DEGENSAC. It uses local optimization to refine estimates and includes degeneracy checks to ensure the resulting matrices are reliable even when the data contains dominant planes or other degenerate structures.

Who it’s for

Computer vision engineers and researchers working on image matching, 3D reconstruction, and two-view geometry estimation.

Highlights

  • High Accuracy: Competitive with state-of-the-art estimators for fundamental matrix estimation.
  • Performance: Often faster and more accurate than vanilla OpenCV RANSAC implementations.
  • Robustness: Implements DEGENSAC to handle degenerate cases that would otherwise bias the result.
  • Cross-Platform: Supports Windows, macOS, and Linux.

Related

  • Project
    OpenStitching/stitching

    A Python package for fast and robust image stitching that creates panoramas from overlapping images using OpenCV.

  • Project
    scikit-tda/ripser.py

    Ripser.py is a Python package that wraps the fast C++ Ripser library to compute persistent homology (Vietoris‑Rips and lower‑star filtrations). It offers functions to get persistence diagrams, plot them, work with images, and retrieve representative cochains, plus a scikit‑learn‑style transformer.

  • Project
    leomariga/pyRANSAC-3D

    An open-source implementation of the RANSAC method for fitting primitive geometric shapes like planes and spheres into 3D point clouds.

  • Project
    triple-mu/YOLOv8-TensorRT

    A high-performance TensorRT acceleration wrapper for YOLOv8, enabling fast inference for detection, segmentation, pose, OBB, and classification in both Python and C++.

  • Project
    cocel-postech/genz-icp

    GenZ-ICP is a degeneracy-robust LiDAR odometry system that uses adaptive weighting to ensure stable localization in environments with few geometric features.