ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Deep Ensemble Network for Movie Scene Boundary Detection
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Jeong-Woo Son, Chang-Uk Kwak, Alex Lee, Minho Han
Issue Date
2017-10
Citation
International Symposium on Advanced Intelligent Systems (ISIS) 2017, pp.1157-1161
Language
English
Type
Conference Paper
Abstract
Movie or video is a mass of semantics. Thus, an analyzing process of movies should be started from segmenting a movie into a set of scenes to reveal precise semantics in the movie. In this paper, we propose a deep architecture to detect scenes in movies with its multiple information. A movie is physically composed of several views that are visual, audio, and textual views. Thus, to handle these different information, it is demanded to construct different models to reflect those information in scene boundary detection. However, it is not so easy to design a single deep architecture for a single view that it cannot be an effective way to construct several architectures to handle multiple views. This paper proposes a method to construct a unified model for all multiple views and we shows its effectiveness with a large movie dataset.
KSP Keywords
Deep architecture, Multiple views, Unified model, scene boundary detection, single view