抄録H-027
ベイジアンネットワークを用いた店舗内の顧客の購買行動認識法における画像特徴に関する検討
ベイジアンネットワークを用いた店舗内の顧客の購買行動認識法における画像特徴に関する検討
○セイ エン・堀 隆之・大谷 淳（早大）
This paper explores the validity of embedding the image processing based extraction of 10 image features that could describe customers’ shopping behaviors to a Bayesian Network based classification method that can recognize the four categories: combinations of with/without hesitation and bought/not bought. By constructing a system that consists of the three modules for tracking customers, extracting the image features and classifying the customers’ behaviors, recognition tests using 48 video data were conducted. As a result of the experiments, a recognition accuracy of higher than 85% is achieved, which is promising.
