Towards Computer-Vision-Based Learning from Demonstration (CVLfD): Chess Piece Recognition

摘要

We present an approach to develop algorithms to offer ‘Learning from Demonstration’. Our aim is to realize Computer Vision as resource-efficient as possible in applications where people interact with computers or -as a special case- with robots. This paper explains the development of a classification program which is to be integrated to a robot that will autonomously play chess. The problem is to perform a classification on a 12 class data set of chess pieces which works on a real-time video feed. We develop two different approaches to solve the problem: A one-step classification is compared to a two-step procedure based on accuracy, computational time and robustness.