Computer Vision Cubed: A Framework for Rubik's Cube Turn Classification

Abstract

The niche world of competitive Rubik's Cube solving is growing. This project seeks to develop an automatic coaching platform for speedcubers' development and efficiency. This task of presenting automatic, targeted feedback to users is broken down into two sub-problems. Leveraging machine learning techniques, this thesis presents solutions to both sub-problems. Due to time constraints, only part of the outlined solutions are successfully implemented in the scope of this thesis. Further work is discussed and proposed.