A framework for semantic classification of scenes using Finite State Machines

We address the problem of classifying scenes from feature films into semantic categories and propose a robust framework for this problem. We propose that the Finite State Machines (FSM) are suitable for detecting and classifying scenes and demonstrate their usage for three types of movie scenes; conversation, suspense and action. Our framework utilizes the structural information of the scenes together with the low and mid-level features. Low level features of video including motion and audio energy and a mid-level feature, face detection, are used in our approach. The transitions of the FSMs are determined by the features of each shot in the scene. Our FSMs have been experimented on over 60 clips and convincing results have been achieved