World Emoji Day is set to take place on 17 July, and Google is taking the opportunity to run an AI experiment disguised as a game. The Emoji Scavenger Hunt is a mini-game that sends players scrambling to match emoji to real world objects. Which is likely a clever way to train some machines to recognise things.
Players are sent on a mission to match 10 emoji with their actual counterparts within a limited about of time. Each item must be found within 20 seconds, although some may be impossible to complete. Like trying to find a bed in the office, or ramen while nowhere near a Japanese store (we’re not sure if Maggi counts).
The scavenger hunt makes use of Google’s Tensorflow.js; allowing it to run completely within the browser. Pretty much demonstrating Google’s ability to get machine learning running on smartphones without the need to send any data to the cloud for processing. We suspect it’s also being used to test the accuracy of Google’s own object recognition system.
Google has also open sourced the code for the game on GitHub.
At the same time, Google has release a list of the most used emoji on Gboard. The most used emoji really shouldn’t be a surprise to anyone. In fact, there aren’t any surprises in the top 10. Leading us to believe that people need to get more creative with their use of emoji.