Catch up on stories from the past week (and beyond) at the Slashdot story archive

typodupeerror

Kinect's AI Breakthrough Explained97

mikejuk writes "Microsoft Research has just published a scientific paper (PDF) and a video showing how the Kinect body tracking algorithm works — it's almost as impressive as some of the uses the Kinect has been put to. This article summarizes how Kinect does it. Quoting: '... What the team did next was to train a type of classifier called a decision forest, i.e. a collection of decision trees. Each tree was trained on a set of features on depth images that were pre-labeled with the target body parts. That is, the decision trees were modified until they gave the correct classification for a particular body part across the test set of images. Training just three trees using 1 million test images took about a day using a 1000-core cluster.'"
This discussion has been archived. No new comments can be posted.

Kinect's AI Breakthrough Explained

on Saturday March 26, 2011 @06:19PM (#35625190) Journal

I don't think so this time. This is a reasonably well written formal paper sent for peer review. It is also quite nice to see this published openly.

• Re:Need a more descriptive summary (Score:2, Informative)

on Saturday March 26, 2011 @08:30PM (#35626096)

The function: f=d(x+u/d(x))-d(x+v/d(x)) would calculate the depth gradient of the pixel. It's possible to reconstruct a three dimensional shape from a 2D image [weizmann.ac.il].

Then your problem is trying to match a human skeleton to the shape. If you know the curvature of the gradient at a particular point, you can eliminate some body parts. A head is mostly spherical and within a particular maximum/minimum, limbs and the torso are more cylindrical with a linear depth along one axis. Look for that linearity, and you could determine that is a limb and what direction it is aligned in.

• Re:Focussing on the normal bit (Score:4, Informative)

<gmaslov@bootis.org> on Saturday March 26, 2011 @09:20PM (#35626404) Homepage

So they fed an LCS with some sample data? OK, par-for-the-course. I'm far more interested in how they generated those '1 million' pre-labelled test images in the first place.

I read the paper; it was clever. They used a standard motion capture setup with their actor(s) going through several hundred different movements. Since their algorithm is stateless, they could analyze the motion and produce many distinct poses from each movement. Each pose was then "retargeted" (a well known technique in animation; example [snu.ac.kr]) onto many different 3D models of people of varying height, body type, etc., before finally being rendered into a perfectly labeled depth map.

They went through several iterations of this process:

1. Train their algorithm on this huge data set
2. Notice that it doesn't work so well in some situations
3. Have their mo-cap actor(s) produce additional data to cover those situations
4. Process the new mo-cap data into however many thousands of additional training poses
5. GOTO 10

Related LinksTop of the: day, week, month.

Were there fewer fools, knaves would starve. - Anonymous

Working...