機械知覚&ロボティクスグループ
中部大学

Human Detection 学術論文(E)

People Detection Based on Co-occurrence of Appearance and Spatio-temporal Features

Author
Yuji Yamauchi, Hironobu Fujiyoshi, Yuji Iwahori, Takeo Kanade
Publication
National Institute of Informatics Transactions on Progress in Informatics, no. 7, pp. 33–42, 2010

This paper presents a method for detecting people based on co-occurrence of appearance and spatio-temporal features. In our method, Histograms of Oriented Gradients (HOG) are used as appearance features, and the results of pixel state analysis are used as spatio-temporal features. The pixel state analysis classifies foreground pixels as either stationary or transient. The appearance and spatio-temporal features are projected into subspaces in order to reduce the dimension of feature vectors by principal component analysis (PCA). The cascade AdaBoost classifier is used to represent the co-occurrence of the appearance and spatio-temporal features. The use of feature co-occurrence, which captures the similarity of appearance, motion, and spatial information within the people class, makes it possible to construct an effective detector. Experimental results show that the performance of our method is about 29.0% better than that of the conventional method.

前の研究 次の研究