Visualization of conceptual model of cooking workspace

Abstract

This paper proposes a method for visualizing and analyzing conceptual models of cooking space in kitchen. We hypothesized that users recognize the space as a set of separated areas to which the users allocate cooking items (tools and foods) along with the concepts in their mind. To model such concepts, we separated cooking areas according to the detected items' location, then characterized these areas by the located items' categories; we use Single shot multibox detector (SSD) for item detection, the EM algorithm with Gaussian Mixture Model (GMM) for dividing the space into areas based on the detected items, and characterize each area with term frequency - inverse document frequency (tf-idf), where we regard each area as a document and items as terms. In conclusion, we visualized their conceptual models of relationships of kitchen items and the areas, and their shape patterns.