ユーザ観察における動作パターンの自動分類によるデザイン課題発見

概要

論文の詳細を見る
In this paper, we propose a technology to automatically detect design problems of the product from user observation videos. In the proposed method, we attach color markers to necessary portions of a user and a product. Then we identify the coordinates of the markers in user observation videos by image analysis and color detection, compare the coordinates among markers qualitatively in every frame of the videos, and classify and encode users postures in all frames of the videos. Through N-gram analysis of qualitative posture sequences in the video, we classify and encode the users behavior patterns as specific sequences of qualitative postures. We assume that the behavior patterns which frequently appear only when the user does not understand the usage of the product should possibly indicate some design problem such as incomprehensibility or difficulty for using. By conducting some experiments to detect such behavior patterns from user observation videos, we show the effectiveness of the proposed method.