Beyond the pixel plane: sensing and learning in 3D
thegradient.pubThis is an exciting time for those of us working on computational geometry to better understand 3D shapes across many industries.
In addition to the architectures mentioned in this great overview, I'm excited to see progress on spectral and geodesic CNNs for graphs and manifolds. Check out this other fantastic source for info on 3D ML: http://geometricdeeplearning.com
This is a great overview. Also checkout CS 468 from Stanford, http://graphics.stanford.edu/courses/cs468-17-spring/ "Machine Learning for 3D Data"
Also, if you want to work on this stuff full time- https://news.ycombinator.com/item?id=17649726
Super cool course! Thanks for the link!
Do you know what are the most precise programmable RGB-D cameras a non-professional can buy? I was trying to extract 3D information just from a single camera via 3D convolutions and RNNs (for a self-driving car project) and would like to play with real 3D a bit as well.
(I wrote the original article)
I've been playing around with a few and I'd recommend the Orbbec Astra and the Intel RealSense (the new D435 is what I've been using) as decent but cheap cameras if you want to get started! The Asus Xtion PRO LIVE is also quite good but since it's been discontinued it's pretty hard to find.
The Stereolabs ZED relies on stereo vision but produces a similar output as traditional RGB-D cameras, and I've heard good things about it as well!
Any idea if the iPhone X surfaces RGBD from the TrueDepth camera?
Yes! Haven’t had a chance to play around with it but I’ve been wanting to. See AVDepthData: docs at https://developer.apple.com/documentation/avfoundation/avdep... and reference implementation for streaming depth at https://developer.apple.com/documentation/avfoundation/camer...
Yes
Thanks for the article. It is well written.
You promise some directions for fruitful research. It is a bit light on that. Maybe it's nice to expand on that topic a bit more.
Not workable for a self-driving car - but for other applications the iPhone X has a front-facing RGB-D sensor.
As an alternative, dumping out pixel ground truth from something like Unreal Engine isn’t hard to do.
What about pose estimation? e.g. Given a well defined coordinate system, like the origin is the nose on a face, determine the pose of the face. Is this still best done with classic optimization formulations like ransac/ICP and a supplied model, or have these been bested by learned models somehow?
Don't think it's exactly what you're talking about (I'm sure there are other works much closer to what you have in mind, just can't recall off the top of my head) — but you might find PoseNet (https://www.cv-foundation.org/openaccess/content_iccv_2015/p...) interesting. Not explicitly 3D, but estimates where in a large-scale scene a picture was taken using an end-to-end convolutional network.
With that said, I think there's still a ton of merit in classical geometric approaches like ICP — there's a real, geometric basis to why they work. Convolutional networks can demonstrate some pretty amazing results, but they're still mostly "black boxes" to us, and a consequence of this is that it's hard to understand why they work and predict when they'll fail. This blog post (by the PoseNet author, actually) articulates the viewpoint well: https://alexgkendall.com/computer_vision/have_we_forgotten_a.... One recent research direction that I personally find really fascinating is designing deep learning architectures around real geometric properties, e.g. as in Skydio's deep stereo work: https://arxiv.org/pdf/1703.04309.pdf
PoseNet on Tensorflow.js does nice head tracking. One can get rough head pose from nose/eyes/ears. but it's crufty.
[1] Web-browser demo: https://storage.googleapis.com/tfjs-models/demos/posenet/cam... [2] Github: https://github.com/tensorflow/tfjs-models/tree/master/posene...
Has any progress been made towards single view 2D -> 3D inference?
yes, a ton! I think the latest exciting work is pixel2mesh: https://arxiv.org/abs/1804.01654 ; can follow citations in their for other relevant recent work.