[dead]
On the visual feedback of the exercises. Does this look better? https://www.loom.com/share/9c90ab0512ff46e0892f77e682545aa2
So there are models for Pose estimation like [PoseNet] or Qualcomm's [MediaPipe] that give you a graph of the body position. This metadata can be converted to a string representation and be fed into a generative model.…
We got some feedback from friends that having a visual clue for the exercises could make a big difference. Something like a video or animation would allow users to follow along without needing to read through the steps.…
[dead]
On the visual feedback of the exercises. Does this look better? https://www.loom.com/share/9c90ab0512ff46e0892f77e682545aa2
So there are models for Pose estimation like [PoseNet] or Qualcomm's [MediaPipe] that give you a graph of the body position. This metadata can be converted to a string representation and be fed into a generative model.…
We got some feedback from friends that having a visual clue for the exercises could make a big difference. Something like a video or animation would allow users to follow along without needing to read through the steps.…