Real Time Food Recognition System using Support Vector Machine and Multiple Kernel Learning

We propose a mobile system for the food recognition, which calculates calorie and nutrition's of food along with this it also records the user eating habits based on timing all the processing of image of food items is performed on a smart phone itself. It's not necessary to send the images to the server to compose a typical smart phone to work in a real-time. In order to perceive the food items we are doing it in three ways, by text, image and timing. To perceive the food items more exactly, we draw out a color scatter diagram and SURF based method then we completely divide into the extra food categories with linear SVM and fast X2 kernel. The user is provided with the facility of saving the food habits according to consumption of food based on timing. We are implementing this system as an android smart phone application for real-time food recognition. In the experiments we archived 75.33% classification rate for top some category. We obtained positive evaluation from used for user for food recording system.