Hierarchical Self-organizing Maps System for Action Classification
Zahra Gharaee, Peter Gärdenfors, Magnus Johnsson · 2017
We present a novel action recognition system that is able to learn how to recognize and classify actions. Our system employs a three-layered hierarchy of Self-Organizing Maps together with a supervised neural network for labelling the actions. We have evaluated our system in an experiments consisting of ten different actions from a publicly available data set. The results are encouraging with 83% correctly classified actions based on the actor’s spatial trajectory.