Understanding Cross-Cultural Visual Food Tastes with Online Recipe Platforms

Abstract

Traces of human behaviour with online recipe portals offer an opportunity to employ a data-driven approach to the study of food culture. Here, we focus on understanding visual aspects of food preference by analysing datasets from China, Germany, and US. Predictive modelling with low-level image features and Deep Neural Network image embeddings show differences in recipe images across datasets and between recipes with high and low appreciation within datasets. Our findings demonstrate the utility of the approach for studying visual aspects food culture.