22 comments

[ 235 ms ] story [ 1466 ms ] thread
Doesn't work for me :( The webcam was clearly on, and it was taking microphone input.
Really hard to navigate. I gave up.
Impressive. I've found that shaking my head was more effective for controlling the slides - but it gives a weird effect...
Didn't work very well. May be the quality of the webcam though
This blew my mind.

It's weird how scrolling forward from your right to your left is more easily detected than scrolling backwards from your left to your right, but it's an amazing proof of concept.

This is pretty amazing – seems to just work. Probably quite dependant on the webcam though.

Remember to move your hand above/below the screen before swiping else you'll swipe back and forwards all the time!

Alternatively, you can move your hand back fast enough for it not to register.
I couldn't get a gesture recognised with my MacBook Air. (Maybe the lighting is a little bad with a window behind me.)
For a proof of concept, this is pretty amazing.
After Leap Motion, I've been very giddy to see where we will soon be in terms of parsing sign language in front of a webcamera.

Hopefully someone finds a financial incentive to pursue that venture. Maybe there's a Duolingo-like idea waiting to be taken advantage of.

Seems to depend on proper skin detection, which can work out of the box if the light is right. Au contraire js-aruco needs preparation, but has some nice 3D capabilities.
There's a bit of confusion on the slides. Mixing up getUserMedia with WebRTC.
It's very impressive that this works at all, although it's extremely inaccurate for me.
My thoughts exactly. Not very usable, but I'm still impressed.
This reminds me Minority Report interface scene.
I really don't get the comparison to Kinect at the beginning, since this is _just_ a simple motion detector and the only fun thing is the execution environment.