Skeleton Detection and Hand Tracking

This project focuses on skeleton detection, particularly for hand and body poses, using various implementations and addressing challenges such as glove detection and real-time performance.

Python

Linux

Academic Research

Pose Estimation

Image Processing

Tensorflow

Project Overview

The project explores pose estimation with a focus on hand and body detection, leveraging existing research papers and their implementations to achieve real-time performance.
Hand Pose Estimation: We adopted the approach presented in Convolutional Pose Machines ↗ paper.
Body Pose Estimation: We built it on the TensorFlow implementation of OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields ↗.
Glove Detection Challenges: Adjusted color masks for detection, successful with different colours.