Camera Calibration using chess board and
Pose Estimation using ArUco tag

This objective of this project was to calculate the camera parameters for pose estimation.

Key Learnings

1. Camera Calibration - Extrinsic and Intrinsic parameters
2. Importance of Fiducial Markers
3. Pose Estimation

Challenges and Workarounds

Getting Roll, Yaw and Pitch from Rotation Vector

Rotation and Translation matrices were used to transform from the world coordinate system to camera coordinate system. Rotation matrix used for this is usually a matrix of dimension = (3, 3). The output (rvec) I got from cv2.calibrateCamera() was a vector of dimension = (1,3). Later I found out that rvec is a Rodrigues vector and we need OpenCV's Rodrigues function to convert this into a rotation matrix of dimension = (3,3)

Conclusion

This task was really fun to do! I also learnt a lot about AruCo Tags and Camera calibration.