On the applicability of the Multi Agent System Paradigm for Parsing Videos