Biological inspired human action recognition

Abstract

Computational neuroscience studies through Functional magnetic resonance imaging (fMRI) claimed that human action recognition in the brain of mammalian pursues two separated pathways in the model, which are specialized for the analysis of motion (optic flow) and form information[3]. For analysis of the form information active basis model is used by different scales and orientations of Gabor wavelets to form a dictionary regarding object recognition (human). In motion pathway, biological movements are recognized by analyzing optic-flow patterns and entering motion information for form pathway adjustment. A synergetic neural network is utilized to generate prototype templates, representing general characteristic of every class. By having predefined templates, classifying performs based on multi-template matching. We successfully apply proposed model on the human action video obtained from KTH human action database as largest human action database. The attained results using proposed methods were promising.