Back in 2016, the “Mannequin Challenge” took over the Internet, with everyone from Hillary Clinton to Beyoncé posting videos of themselves standing still in various scenarios. Google is now using these videos to train their AI.
One of the top challenges with AI is teaching machines how to move through unfamiliar surroundings. By training machines to interpret 2D videos as 3D scenes, it can help them understand the depth of a 2D image. Google has decided to leverage the thousands of YouTube videos of the Mannequin Challenge as a data source to help machines understand depth since the videos show people posing from all different angles.
Why It’s Hot
These learnings will be particularly useful in AI for self-driving cars.