Show HN: Incremental SfM pipeline that reconstructs 3D point clouds from images

Category: library

Tags: structure-from-motion, 3d-reconstruction, computer-vision

Score: 5.3/10 (Innovation: 4, Technical: 6, Documentation: 7, Utility: 4)

An incremental Structure from Motion (SfM) pipeline implemented in Jupyter Notebook that reconstructs sparse 3D point clouds and camera trajectories from ordered image sequences, demonstrated on a medieval tower. It provides a clear, step-by-step explanation of SIFT feature matching, essential matrix decomposition, triangulation, and incremental PnP registration, making it educational for computer vision enthusiasts.

Target audience: computer vision researchers, data scientists, hobbyists

Repository: https://github.com/egeozgul/Incremental-3D-Reconstruction-SfM/tree/main · Jupyter Notebook

View on Hacker News