This project aims to reconstruct a hand in a 3D viertual world from the video footage from a webcam. This can be used for several purposes like more interactive games, interactions with virtual world with our own hands, etc.
This uses Google's MediaPipe Hand Landmarker for extracting the data about the hand from the image frames of the video footage.
It uses OpenCV to get a direct video feed from the webcam and preprocess it to a bit, then it passes the data to the mediapipe hand landmarker model which gives us the list of detected landmarks, which is then filtered and transformed by a custom a algorithm.
The final positions and labels are then passed to the CGL core for being displayed inside the 3D world!