A machine learning assistant for choosing operators and tuning their parameters in image processing tasks

Abstract

Operator choosing and parameter tuning in image processing tasks is still to be a challenging problem to achieve good results. This paper discusses the formulation of a solution combining both machine learning and multi-agent system to help users to choose best operators and attribute appropriate values to their parameters in image processing applications. The aspect of cooperative learning makes this solution faster and outperforms one agent learning. The empirical study shows that our solution is effective on finding the optimal vision operators and their parameters’ values.