Comparison of Various Classifiers over Type I and Type II Diabetic Food Recognition System

The inability to control the infection of diabetic people, computer-aided automatic food detection system has wedged more attention now days. The food image processing is the most gifted tool is used for food identification. The key point extraction's Scale Invariant Feature Transform (SIFT) algorithm is used to extract the key points from food image, which is used for building visual dictionary of visual words based on color using k-means clustering algorithm. Features can be grouped into separate classes, namely class I and class II using multi-label Support Vector Machine (SVM) classifiers such as SVMlinear, SVMrbf, SVMexp, PARZEN WINDOW and K-Nearest Neighbor (KNN) to identify the input image belongs to which class. Class I have diabetic patient's eatable food images and Class II have diabetic patient's not eatable food images. GLCM use contrast, correlation, energy and homogeneity parameters to measure various calories from food image for diabetic patients. Finally measure and compare the recognition accuracy for various classifiers. The recognition accuracy for various classifiers is used to prove the feasibility of the approach in a very huge food image dataset. This project is about consciousness on food particularly for diabetic patients. citation: Anusha B, Regional Center Of Anna University, Tirunelveli.; Ashin Leo ,Regional Center Of Anna University, Tirunelveli.. "Comparison of Various Classifiers over Type I and Type II Diabetic Food Recognition System." Global Research and Development Journal For Engineering 16 2016: 48 - 57.