Last night at the Creative AI Meetup we finally got to have people dancing and our deep learning model responding!
We built a model that takes in a given dance (perhaps captured via OpenPose), and responds with it’s own interpretation of that dance, done in it’s own style. Sometimes the computer responds with an entire dance crew, where each individual dancer in the dance crew is responding to the same input dances, but dances it in their own unique style! You’ll note some wackiness in the videos occasionally, and that’s mostly because the network responds best when there is an entire single person in the frame, if it doesn’t see that, or if multiple people pop in and out, it tends to get confused. We’ll be refining these aspects over time.
You can watch the videos below, and you can grab the JSON data for the responses here.
You can find my talk here where some high-level detail of the model is covered.
Craziest dance of the night
Looking forward to doing more with this over the coming months!