A novel approach to detect anomalous behaviour using gesture recognition

Abstract

The primary goal of my work is to create a system which can identify specific human gestures and group the common gestures which in turn is used to convey information if uncommon activities are performed. Identification will be based on a video input based self learning gesture identification model which will classify the gestures based on genetic parameters. Most papers in this area focus on classifying different gestures, but do not judge whether the recognized gesture is good or bad in continuous recordings of daily life. The uniqueness of my approach lies in the method to manage a process of mass gesture detection in common places and classifying it using Support Vector Machine in the Learning mode. In the Execution mode, video footages are fed to my model which compares the current patterns with the stored normalized patterns and flag the ones that are odd. © 2013 Springer.