“Teaching Google Photoshop” has been my working mantra here—i.e. getting computers to see like artists & wield their tools. A lot of that hinges upon understanding the shape & movements of the human body. Along those lines, my Google Research teammates Tyler Zhu, George Papandreou, and co. are doing cool work to estimate human poses in video. Check out the demo below, and see their poster and paper for more details.
3 thoughts on “A cool demo of Google’s pose estimation tech”
I wonder if a variation of this technology could be used to help photographers help their models pose for whole body portraits, at shooting time, using some kind of standard posing template viewfinder overlays.
I like that idea, but with a slight variation: “Normal” people are often uncomfortable posing for a picture. Where do I put my hands?, etc. How about a camera that tracks the subject’s pose as they move around and automatically snaps the shutter when the subject is an an “aesthetically pleasing” pose.
Cool idea! It’s possible to automatically fire the shutter when people are smiling, have their eyes open, etc. Extending that to poses would be interesting.