Visualization Based on Person Move Similarity for Person Behavior Analysis
Hidekazu Yanagimoto, Kiyota Hashimoto, Tokuro Matsuo · 2019
This paper proposes new visualization based on distance between persons, which are defined with Normalized Restricted Dynamic Time Warping. In the visualization persons are located based on their distances that are defined by their movement histories with t-SNE. In experiments we visualize academic conference participants with our proposed method and confirm that the proposed method helps us to find participants moving together.