Emoji Analysis Using Deep Learning

Abstract

Emojis or avatars have been the ways for indicating nonverbal cues. These cues have become the most essential part of online chatting, product review, brand emotion, and so on. It has been leading to increasing data science research dedicated to emoji-driven storytelling. With advancements in computer vision and deep learning, it has been currently possible to detect human emotions from various images. In this deep learning project, we have classified human facial expressions to filter and mapped corresponding emojis or avatars. In 1999, Japanese artist Shigetaka Kurita created the first emoji. He worked with the people, who worked for a renowned mobile Internet platform, named ‘I-mode’ from DOCOMO which was Japan’s operator company for the mobile phones. The original 176 emojis introduced by Kurita are currently among the persistent collection at Museum of Modern Art which is situated in New York, which are the confidential symbols over faces, because it was DOCOMO’s objective and motive to search for better ways to communicate and convey ourselves. Emoji became a boon and sense of popularity in Japan very quickly. On the other hand, DOCOMO’s idea was being taken by other rival mobile companies.