Emoji Scavenger Hunt

Objective

Google Brand Studio created this interactive experiment to showcase the practical applications of neural networks and machine learning to a broad audience. The project aimed to make AI technology tangible and engaging by enabling users to experience real-time object recognition through their smartphone cameras in a playful, accessible format.

Solution

An interactive browser-based game built on TensorFlow.js where players received emoji prompts and used their phone's camera to find matching real-world objects. The application ran entirely in-browser without server connections, ensuring privacy—no camera images were stored or transmitted. The game employed machine learning models to identify whether camera-captured objects matched target emojis. Players started with a 20-second timer that extended with each successful match, creating a timed challenge mechanic. The experience was distributed through the Experiments with Google platform (g.co/emojiscavengerhunt) and made open-source on GitHub.