In this paper, we propose a method for human detection
using co-occurrence of Histograms of Oriented
Gradients (HOG) features and color features. This
method expresses the co-occurrence between HOG
and color features by Adaboost and generates the
combination of the features effective for the identification
automatically. Color features were calculated
by making histograms that quantized hue and saturation
in local areas. We show the effectiveness of
the proposed method by identification experiments
for human and non-human images.