Accuracy Analysis for Image Classification and Identification of Nutritional Values Using Convolutional Neural Networks in Comparison with Logistic Regression Model

Abstract

The goal is to raise public awareness about nutritional issues by using food images to predict nutritional analysis using a novel image classification technique. Methods and Materials: The proposed research will be conducted at our university, and a total of two groups have been formed. There are two types of neural networks: a convolutional neural network and a logistic regression network. The framework uses 10 samples per group to evaluate accuracy. Gpower of 80% was used to calculate the sample size. Results: Convolutional Neural Network algorithm has predicted the nutritional analysis with the accuracy of (83.84%) which is more compared with the Logistic algorithm (72.3%) in identifying the fruit, Calorie count, amount of protein content, total fat, and subsequently carbohydrates measurement and so on. There is no statistically significant difference with (P = 0.092, >.05) among the classification algorithms. Conclusion: The analysis shows that the Convolutional Neural Network is significantly better for the whole Nutrition Analysis process compared to the Logistic regression.