Twitch and other live stream platform has made it for gamers who has high latency and high bandwidth, but for fitness professionals, yoga teachers and dance instructors that's still a high barrier of entry. At AT&T 5G Hackathon we were able to utilize 5G and AI on the edge to empower the streamers to broadcast everywhere. With AT&T Next Generation TV, we can now enable the viewers to consume the content that's provided by the streamers..

What it does

The fitness professional and yoga instructors now can stream their class live, the current platforms are not interactive and lacks feedback loop, the effect makes no difference than simple video recording. With help of AI on the edge we can provide feedback loop to the streamers so they can be more aware of their class. The viewers now can consume the content with their USB camera on top of their AT&T Next Generation TV. We can utilize the AI on the Edge on the AT&T Next Generation TV to empower entire industry of fitness streamers and making these classes a lot more accessible with comfort of your home.

How I built it

We built an android application on AT&T TV, connecting with USB camera and using tensorflow to get the human skeleton tracking. The skeletal tracking returns to streamers at feedback loop so that the streamer knows how much attention people are paying.

Challenges I ran into

Getting USB Camera running on Android TV was a pain, as it's not considered camera. Getting AI on the Edge running on TV was also extremely difficult, it does not see usb camera as part of android camera system, it took a long time to get it done properly

Accomplishments that I'm proud of

I was able to complete the application that utilizes both camera and AI on the Edge on the TV device. But the most important part, I was able to impressed many yoga teachers with the idea and prototype, who are all want to become fitstreamer when we launch

What I learned

Connecting USB camera on Android TV, getting AI on the Edge and Tensorflow to run on Android TV.

What's next for FitStream TV

We've already started to do user research and talking to different yoga teachers, our next step is to sign up as many yoga teachers as possible before we launch the platform.

Built With

  • ai
  • android-tv
  • att-tv
  • tensorflow
Share this project: