Skip to Main Content
Human action recognition and interpretation constitutes an important part of the video understanding. In this work, a novel action recognition system is developed that uses edge features obtained from optical flow power shapes which is represented as sequential gradient histograms. The presented system can achieve equal results to the complicated top action recognition systems of nowadays. The system is tested with the Weizmann dataset which is widely used in the field, and comparisons are given.
Date of Conference: 18-20 April 2012