In order to understand the etiology of obesity related to people's diet, we have developed a method to calculate food volume (portion size) from a single digital image in which a dining plate was used as a reference. In order to evaluate the error of this method, we observe the relationship between the relative error under different food imaging scenarios, such as angles of camera rotation and distances between food and camera. Based on our result, an error reduction scheme is proposed to improve the accuracy in image-based food volume estimation.
展开▼