Object Pose Estimation using OpenCV

This repository consists of collection of tasks on object detection and pose estimation using OpenCV.

Task 1

PnP Algorithm for Object Pose Estimation

PnP.py pnp.gif

Finds the pose of the box using the PnP algorithm from the vertices of the box.

Task 2

Establishing correspondences for SIFT Points

sift_matching.py sift_matching.gif

Matches the 2D and 3D correspondences of the SIFT points using the Ray-Box intersection method

Task 3

RANSAC Algorithm for Object Pose Estimation

ransac.py ransac.gif

Uses the RANSAC algorithm to find the object pose from the 2D and 3D correspondences found in the previous task.

Task 4

Camera Trajectory Estimation by minimising the Reprojection Error

reprojection_error_initial_pose.py reprojection_error.py reprojection_pose.gif

Finds the pose of the object by minimising the reprojection error of the SIFT points between subsequent object images.

Acknowledgements

The tasks and pictures are based on the Tracking and Detection for Computer Vision course offered at TUM.