Action Recognition and Suspicious Action Detection with Mixture Distributions of Action Primitives

Abstract

In this paper, we propose a generic framework for detecting suspicious actions with mixture distributions of action primitives, of which collection represents human actions. The framework is based on Bayesian approach and the calculation is performed by Sequential Monte Carlo method, also known as Particle filter. Sequential Monte Carlo is used to approximate the distributions for fast calculation, but it tends to converge one local minimum. We solve that problem by using mixture distributions of action primitives. By this approach, the system can recognize people's actions as whether suspicious actions or not.