Analysis of Food Calorie using Extended Kalman Algorithm and Novel Hybrid Deep-Learning Framework

Food calorie detection forms to be the crucial aspect of an individual's quotidian life. With the rapid escalation of diseases, it is cardinal for an individual to explicitly manage their diet, and pivot toward a healthy lifestyle. Food images have been instrumental in rendering large amount of information pertaining the calories they hold, and thus aid in balancing the intake of a person. Inorder to obtain the pinnacle of accuracy in obtaining the calorie of the food to be consumed. This study focuses to enhance the image quality of processing food, along with comprehending the detection mechanism used to analyse the impact the food would have on the body. The incorporation of gaussian noise elimination as pre-processing mechanism to ameliorate the image quality, along with CNN and Kalman filter algorithm is implemented to train, test and predict the calories present in the food. A phase of verity used in this study is the entailing of regression to effectively evaluate the actual versus the predicted calories obtained from the food images. The results for this indagation are procured through MATLAB, and is successful in establishing the novelty through the novel hybrid layers effectuated with CNN, inorder to augment the wholistic efficacy of the research.