Project using PoseNet & KNN

Tips before trying:

- Kindly restrict your poses (don’t go too wild!)
- If you are not satisfied with a position, click “Create” and repose in a different spot :)
- You can always refresh the page and record new poses in for different representations!

Try it here!

Demo Video

Post-production Statement

In general, my project is working towards the original artist statement I wrote: the users are allowed to create drawing elements with their poses on a blank canvas. But what I’m lacking compared with the original statement was having the user be part of the world they create. My original idea was that a human character indicating the user will appear on the canvas also, and the character can travel on the canvas when the user leans to different sides. I still think that idea is achievable by giving it a large canvas that moves when the user leans. It can even be better than how the project is positioning the drawing elements now because I can fix the character itself in the middle of the screen and indicate its movement by having it facing different directions. When the “create” button is pressed, the element will appear on the same x-coordinate (which is always the middle of the screen in this case), but will be in various positions relative to the canvas. I’m also thinking about having a better way to trigger the create button instead of clicking it. Even though a certain period of time is provided before the model starts to classify, the user still has to come near the screen and click the button and move to another position to pose. Maybe I can implement a certain pose that indicates “create” so that when that pose is made, it means the button is clicked and the user can continue drawing without having to move closer every time.

github: https://github.com/jamiewang76/poseCreate
Interactive Art Projects