Deep Learning Driven Food Recognition and Calorie Estimation Using Mobile Net Architecture

Obesity is a serious disease and it is on rise in now a days because food can be easily brought to our place without any efforts. As a result, the need for food increases and at same time curiosity regarding the nutritional content will also increases. As the food availability becomes easy along with that lots of health issues are also recorded due to overeating, so people become aware about their accurate diet plan to protect themselves from diseases which are caused due to obesity. The main aspect in any of the diet plan is nothing but the calories content within the food item. If the people accurately know the caloric information of the food they intake then people become cautious and tries to avoid it to protect their health. So to solve this problem we proposed a deep learning-based technique to identify the calorie content present in the food image given by the user. This project promotes healthy eating habits through deep learning techniques. The primary goal of the project is to accurately identify different food items and provide real time predictions of the calories content. The proposed system leverages a Mobile Net-based convolutional neural network (CNN), trained on a diverse and extensive of food images, to accurately recognize an extensive range of food items. The MobileNet architecture, known for its efficiency and adaptability, is employed to enable real-time food identification and maintaining greater accuracy levels with less parameters. Moreover, the system incorporates accurate calorie estimation module that utilizes identified food items to estimate nutritional values.