Classification for Myanmar Food Image using DCNN

In this paper, Myanmar food images are classified using a deep convolutional neural network (DCNN). Food images vary greatly even within the same genre, making it challenging to classify them. A convolutional neural network extracts spatial features from images, making it highly efficient for solving image classification problems. Thus, it has become popular practice to do research on Convolutional Neural Networks (CNNs) for food classification using food images. In this study a dedicated deep model has been developed for the extraction of features and classification of Myanmar food. In the future, this system can assist in monitoring individuals' daily calorie intake and tracking their eating habits through camera-based food consumption tracking. A newly developed Myanmar food dataset (MF15 dataset) includes fifteen categories, such as Ei Kyar Kway, Shan Noodle, and etc. The dataset contained about five hundred images for each class, totaling 7500 images. The proposed method achieved an overall accuracy of 88%. According to the comparison with AlexNet model, the proposed model outperforms AlexNet on the MF15 dataset.