Structure from Motion (SfM)

This Structure from Motion (SfM) project reconstructs 3D scenes from multiple images. The pipeline detects and matches image features, triangulates them to estimate 3D points and camera poses, and optimizes the results using bundle adjustment. The final 3D structure is visualized with OpenCV’s cv::viz, displaying a refined point cloud and camera orientations. This project integrates CUDA, OpenCV, Eigen, and g2o for efficient processing and accurate 3D modeling from 2D images.