IMPLEMENTATION OF NIGERIAN INDIGENOUS FOOD IMAGE RECOGNITION SYSTEM
F. A. Ajala, Folowosele A.O, Jeremiah Y.S, Atanda O.G, Adigun E.B., Abdulkareem Q.B · International Journal of Software & Hardware Research in Engineering · 2020
Food Image Recognition is one of the promising applications of visual object recognition in computer vision.A couple of Nigerian indigenous mealsare on the verge of going extinct in the near future, it is imperative to work towards preserving the knowledge of these indigenous meals.The aim of this study is to implement a Nigerian Indigenous Food Image Recognition System.In this study, a dataset consisting of12 categories and 400 images of indigenous Nigerian meals was downloaded from Kaggle.com, the images were pre-processed using the median filter and Gray-Level co-occurrence matrix, then, the features were extracted and classified using the Convolutional Neural Network algorithm.A 73% level of accuracy of correct recognition was achieved with the model.Based on our findings, Convolutional Neural Network has a higher level of accuracy than other traditional algorithms in automatically segmenting and extracting features and providing accurate classification.