Improved scene classification using efficient low-level features and semantic cues

Prior research in scene classi cation has focused on mapping a set of classic low-level vision features to semantically meaningful categories using a classi er engine. In this paper, we propose improving the established paradigm by using a simpli ed low-level feature set to predict multiple semantic scene attributes that are integrated probabilistically to obtain a nal indoor/outdoor scene classi cation. An initial indoor/outdoor prediction is obtained by classifying computationally e"cient, low-dimensional color and wavelet texture features using support vector machines. Similar low-level features can also be used to explicitly predict the presence of semantic features including grass and sky. The semantic scene attributes are then integrated using a Bayesian network designed for improved indoor/outdoor scene classi cation. ? 2004 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.