3 ms·
I think the problem you'll face is that there are lots of ambiguous inputs: the same pixels could be an image of something small close up or something large fur
by vilya 15y ago
I think the problem you'll face is that there are lots of ambiguous inputs: the same pixels could be an image of something small close up or something large further off, for example. That's likely to confound the training process.
The kinect dodges this problem by actively gathering extra info. It projects a rapidly changing pattern of infrared light out in front of it; based on the pattern it detects it can figure out how long the light took to reach a reflective surface at each pixel and infer the depth from that.
- marshallp 15y agoOf course, it will be less accurate than the kinect itself (but spreading software is cheaper than hardware so it's worth it.). One of andrew ng's students trained this knd of system (3d from images trained using 3d data) a few years ago and was accurate to automatically drive a toy car. That was using just a limited amount of data. Using data from thousands of kinect users will better results than that. It's all a matter of data, the more data you have, the more accurate you can be.
- kabdib 15y agoThe difference between something in a lab and a commercial product -- or even something that is moderately usable in uncontrolled conditions -- is often vastly underestimated. The effort to make something real is probably several orders of magnitude harder than what's required for "I can hack something up in a couple of weeks."
- marshallp 15y agoI believe it has more to do with how much data you can get your hands on. With the kinect - widely available depth data, and cloud computing, I believe a lot of things that seemed difficult are now possible. Open source movement needs to start shifting to crowd sourced data and volunteered compute time movement.
- vilya 15y agoA large publicly available depth image dataset could be quite valuable as a research tool.