Publication
People's liking for cooked food is affected by a number of factors, including appearance, taste, smell, and eating habits. Among all these factors, appearance plays a vital role, especially in some situations where only the appearance of food is available on mobile displays. However, previous research on the effects of appearance on people's liking for cooked food is limited in dimension and scale. In this paper, we investigate the relationship between three major visual aspects of cooked food and their visual appeal. We propose and extract several visual features in terms of color, texture, and layout, based on images collected from a large online food community. We also train classifiers using proposed visual features to predict the visual appeal of cooked foods. The results show that we can achieve about 77% prediction accuracy, and we find people prefer cooked food with bright and warm colors, and a smooth surface.