Built a lightweight GUI demo to perform human pose estimation on video using MediaPipe’s pretrained model. Integrated OpenCV for visualization and playback control. Allowed display rate adjustment to simulate processing delay and visualize frame-by-frame pose prediction results.

Pose Estimation with OpenCV

This project demonstrates a basic implementation of stereo pose estimation using OpenCV in Python. It includes:

• Keypoint detection & matching
• Fundamental & Essential matrix computation
• Camera pose recovery via cv2.recoverPose
• Triangulation of 3D points from stereo pairs

Goal: To gain hands-on understanding of epipolar geometry and visual localization.

Reference: https://www.youtube.com/watch?v=I1gMUbEAUFw

===========================================================================

You must install opencv-python(cv2) and mediapipe before excuting this module.

By the way, the python devolopment is ver3.9

===========================================================================

Usage:

1. Put this module in any dirctory.
2. Put any media file (.mp4, .gif, etc.) in the same directory
3. Excute this module in cmd: python poseesti.py
4. Follow the instruction in the program: (a) input the file name (including the file extension) (b) input the display rate (recommend 70+)

Enjoy!