Calorify: Calorie Estimation System Using Object Detection by Deep Learning Technique

Abstract

In last decade or two, an incremental growth in obesity has been witnessed all around the world. There has been increasing research in order to tackle obesity using food logging and  food image calorie analysis. These researches allow users to monitor their calorie intake which helps promote healthy living. This paper explores the use of image analysis techniques to ascertain a more accurate calorie count from photographs and images of food items. The methods employed involves determining the calorie count of the food item through mathematical calculations of the features extracted from food image by using deep neural networks. In this paper, we propose a mobile software for food calorie estimation from pictures of food plates. By using YOLO [6] (You Only Look Once- Object detection using deep learning algorithm), we are able to detect the food and thereby calculate the required food calories from the diverse datasets of Indian cuisine.