An Approach to Food Recognition and Nutrition Assessment Based on Machine Learning

Abstract

To prevent obesity in the human body, a balanced diet is now required for regular consumption of healthful foods. This research, introduced a novel machine learning-based system that automatically classifies food photos accurately and calculates food qualities. In the training portion of the prototype system, the deep learning model proposed in this paper uses a convolutional neural network to classify food into several categories. Increasing the pre-training model's accuracy is the primary goal of the suggested approach. A client-server model-based prototype system is designed in this paper. An image detection request is sent by the client, and the server handles it. Three primary software components make up the prototype system's design: a server-side module, a text data training module for attribute estimation models, and a pre-trained CNN model training module for classification applications. To attain better categorization accuracy, made experiments with several food categories, each with hundreds of photos, and machine learning training.