Android based application for recognition of Indonesian restaurant menus using convolution neural network

摘要

Indonesia has a lot of typical food which is rich in flavor. However, compared with countries such as China, Thailand or India, Indonesian food is not widely known by the international community. One of the factors that cause Indonesian food is less popular is that the food information cannot be accessed easily by foreign travelers during a visit to Indonesia. Food name is written on the restaurant menu can be read, but not really showing the information about the food, such as ingredients, how the food is prepared or a picture of the food itself. Hence, it is required to facilitate foreign travelers who want to find out information about Indonesian food based on restaurant menus. An Android-based application has been developed to show food information. The application captures a text on the restaurant menus, process and recognize the text using convolution neural network (CNN). The recognized text is then matched with predetermined database to show the information about the food. The application was able to recognize 100% of the menus when the menus use Sans Serif Font. However, the accuracy dropped into 56% when the menus use Times New Roman.