IMPLEMENTATION OF CONVOLUTIONAL NEURAL NETWORK TO CLASSIFY HUMAN GENERATED DOODLES IN A MOBILE GAME ON ANDROID OS

Abstract

Artificial intelligence has proven its benefit in adoption when incorporated with any type of processes. Various fields and complexities such as science, military, health, economy, entertainment, and so on have developed artificial intelligence with unique purposes. May it be for assistance or full control, tasks can be completed with less effort. Artificial intelligence has been gaining traction exponentially in the past years, especially in business and science fields. However, educations on artificial intelligence is relatively less developed. The idea of familiarizing children early on to the latest technology trends tend to be beneficial in their lives later on. Other than that, having young contributors could bring fresh and revolutionary ideas into the research and development of artificial intelligence itself. This project will develop a doodling mobile game utilizing a convolutional neural network, to engage the audience into a deeper understanding on artificial intelligence. It features a real-time predicting artificial intelligence (AI) model while the user doodles, the AI tries to guess what keyword the user is doodling. Other than that, it also features an animated doodling demonstration for previously drawn doodles along with the predicted keywords chronologically, so that the user can see in the decision-making of the AI in real-time. Furthermore, the doodles that had been drawn will be used as further improvement to the AI in order to achieve better performance in its processes.