Food Detection and Calorie Estimation of Food and Beverages using Deep Learning

People nowadays, are not maintaining a healthy body due to intake of unnecessary and unhealthy diet. Hence, it is required in recent times, to maintain their standard diet so they can avoid various problems related to obesity that cause risk to their healthy body. So in this paper, we are presenting an automatic machine learning-based novel system that classifies the food items, identifies them, and estimates the calories present in them i.e. estimation of calories is performed. We have obtained an accuracy of 89.48%. For the classification of the food images and identification, we suggest using a Convolutional Neural Network based on a deep learning model that will perform the required purpose in the training part of the system. The proposed model would utilize a client- server architecture, wherein the client uploads an image. A needed algorithm on the server side then executes to assist in estimating the caloric content of the corresponding food items. Here we are using the dataset of thousands of images from Kaggle and applying the appropriate algorithm to predict the estimation with the best accuracy. Here we will be using CNN and then pre- trained CNN models such as MobileNet, ResNet, etc.