Machine Learning-Based Approach to Food Recognition and Nutrition Estimation

Nowadays, a consistent intake of healthful foods is required to maintain a balanced diet and to avoid obesity in humans. In this research paper, we introduce a unique machine learning-based system that automatically performs accurate food image recognition and predicts food qualities. In the training stage of the prototype system, this research proposes a deep learning model consisting of a convolutional neural network that classifies food into specified categories. The proposed method's major goal is to increase the pre-training model's accuracy. The paper proposes a client-server model for a prototype system. The client sends an image detection request, which the server processes. The prototype system consists of three main software modules: a pre-trained CNN model training module for classification, a text data training module for attribute estimate models, and a server-side module. We have done experiments with a range of food options, each with hundreds of photos, in order to improve the categorization accuracy through a machine learning training system.