[4J1-OS-25a] 食とAI(1/2)

[4J1-OS-25a-02] Acquisition of Dish-Cuisine Relations Based on the Characteristics of Food Recipes

Hayate Fukumoto¹, 〇China Takahashi¹, Mitsunori Matsushita¹, Ryosuke Yamanishi¹ (1. KANSAI UNIVERSITY)

Keywords:Multimedia on Cooking, Cooking Recipe, Dish Dataset

The final goal of this research is to develop a recommendation system of a plate for dish. When serving food, it is necessary to consider the shape and appearance of the plate. We aim to support the selection of plates for cooking beginners. So far, we have constructed a plate data-set that stores the physical characteristics of plates; however, it is still not enough for practical use. Information about the kind of foods that can be served on plates is necessary to recommend plates. In the proposed method of this paper, we used a feature vector from ingredients and cooking procedures included in the recipe. The vectors were used as input to a three-layer neural network to estimate the attributes of the plate. As a result, we found that the accuracy of the recipe feature vector using both ingredients and cooking actions was higher than other types of features.