Semantic film preview classification using low-level computable features

Abstract

This paper presents a framework for the classification of feature films into genres, based on computable visual cues. The authors view the work as a step towards high-level semantic film interpretation, currently using low-level video features and knowledge of ubiquitous cinematic practices. Our current domain of study is the film preview (the commercial advertisements primarily created to attract audiences). A preview often emphasizes the theme of a film and hence provides suitable information for classification. In our approach, we classify movies into four broad categories: Comedies, Action Films, Dramas or Horror Films. Computable video features are combined in a framework with cinematic principles to provide a mapping to the four high-level semantic classes. An unsupervised clustering technique is used to discover the structure of the mapping between the computed features and each film genre. Through experiments, we notably demonstrate the structure that exists between low-level features and high-level film genres.