Analysis of Convolutional Neural Networks on Indian food detection and estimation of calories. (2022)

• Record Type: Journal Article
• Authors: Sathish, Suriyakrishnan; Ashwin, S.; Abdul Quadir, Md.; Pavithra, L.K.
• Is Part Of: Materials today. Volume 62:Part 7(2022)
• Page Start: 4665
• Page End: 4670
• Publication Date: 2022
• Subjects: Food detection -- Deep learning -- Calorie estimation -- OpenCV -- Indian cuisine; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2022.03.122
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: Indian Cuisine has a peculiar aroma and flavour distinct from other cuisines. On the other hand, Obesity, Diabetes, and Hypercholesterolemia are severe problems in the Republic of India. This research aims to develop and analyze a Deep Learning model based on OpenCV for identifying Indian cuisine and determining their calories. Indian Cuisine dataset was built by extracting images from the internet and preprocessing them to fine quality. From the dataset, the classification of Indian Cuisine was done by Convolutional Neural Network, and with the help of image processing techniques, calories of the classified food were calculated. After an enormous amount of analysis, the developed model detects Indian food cuisine with an accuracy of 99.19% on training data and 95.30% on testing data; also, the estimation of calories was the accuracy with an error variation of ±10 calories to the actual food.