bit.studio as part of YouTube and Google team, has created a new dance experience for YouTube FanFest 2O2O.
We developed a software with a machine learning model that allows users to enjoy their dance movement and get their own dance video with Matt Staffanina, the most followed dance creators on YouTube. Users can download their dance syncing videos with Matt and share it out.
Do you want to be part of the dance movement with Matt Steffanina?
What you need to do is very simple. You learn the full dance in the tutorial video, then try and keep up your move with Matt. When ready, record your dance movement. And our ML model will detect your movement and segment your body out. Then, feature the moves on your dance video!
Try syncing your dance, share it out, and challenge your friends here!
Scalability is very important in this project since we need to support millions of users concurrently. We decided to run our ML model on Google Kubernetes Engine with dedicated graphic cards. This way we can generate the videos with just a short amount of time.
We had so much fun creating such an enjoyable experience for YouTube FanFest 2O2O. Now it’s your turn! Let’s dance! → yt.be/ytffdance
