Slashdot Mirror
Kinect's AI Breakthrough Explained
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.'"
Where can I buy one of those? Kinect is amazing and all but not having to wait 2-minutes for windows to load would be even more incredible! It's funny, computers get more and more powerful but yet windows still takes the same amount of time to load. lulz
First Post!
This is actually pretty brilliant.
cnn emulates real life for US? babys et al differ
hard to tell which smells worse? the fog of tax free (for some) war can do that? did we say tax free? pardon, the non-taxpayers actually profit ($billionerrors$) on the heavy weapon (keeping ALL sides supplied including mexico) murder massacre business outings. so that's good?
we support the views of this former person
http://www.youtube.com/watch?v=TY2DKzastu8&NR=1&feature=fvwp ("stop killing")
we do not support the material in this cnn propaganda video from yesterday
http://www.youtube.com/watch?v=BXB75IK6pL4 ("we can win this, with my help")
same guy? clone? confused? schizophrenic? we must focus... on the images. we must....saw a picture of one of those godaffy psycho-killer freaks being paraded around our military bases (may have paid for them, along with our holycost tithing's) like royalty, only to become our very worst 'enemy' just weaks/leaks later? focus-pocus?
babys rule, with tiny chubby soft fingers, advanced dna etc..... unclear?
real math; taking one (1) life =crime vs. humanity
give US a minute here. this can't be right? isn't there justdenyable homicide? that's the old time religion? god's will? too many of us (by about 5 billion)? still foggy? in these complex times, it can be disgustingly enlightening to return to the teachings of the georgia stone trustdead freemason 'math'?
freemason kids traumatized by native teachings
we're not the only (chosen) ones? the natives must have made some mathematical errors? let's see, wasn't that problem taken care of before? & before that. let's check the georgia stone, all the answers are there? not to fret then, the #s never lie?
the GSM get their tiny (ie; selfish, stingy, eugenatic, fake math) .5
billion remaining population, & the money/weapons/vaccine/deception/fake
'weather' alchemist/genetically altered nazi mutant goon exchangers, get
us? yikes
the 'fog' is lifting? more chariots will be needed?
with real math, even being remotely involved in lifetaking (paying for, supplying endless ordinance) is also a crime against ALL of the world.
ALL (uninfactdead) MOMMYS......
the georgia stone remains uneditable? gad zooks. are there no chisels?
previous math discardead; 1+1 extrapolated (Score:mynutswon; no such thing as one too many here)
deepends on how you interpret it. georgia stone freemason 'math'; the .5 billion, then
variables & totals are objective oriented; oranges: 1+1= not enough,
somebody's gotta die. people; 1+1=2, until you get to
1+1=2 too many, or, unless, & this is what always happens, they breed
uncontrolled, naturally (like monkeys), then, 1+1=could easily result in
millions of non-approved, hoardsplitting spawn. see the dilemma? can
'math', or man'kind' stand even one more League of Smelly Infants being
born?
there are alternative equations being proffered. the deities (god, allah,
yahweh, buddha, & all their supporting castes) state in their manuals that
we needn't trouble ourselves with thinning the population, or being so
afraid as to need to hoard stuff/steal everything. chosen people? chosen
for what? to live instead of us? in the case of life, more is always
better. unassailable perfect math. see you at the play-dates, georgia
stone editing(s) etc... babys rule.
exploding babys; corepirate nazis to be caged (Score:mynutswon; hanging is too good for them?)
there are plans to put them, (the genetically, surgically & chemically
altered coreprate nazi mutant fear/death mongerers (aka47; eugenatics,
weapons peddlers, kings/minions, adrians, freemasons etc...)) on display
in glass cages, around the world, so that we can remember not to forget...
again, what can happen, based on greed/fear/ego stoking deception.
viewing/feeding will be rationed based on how many more of the creators'
innocents are damaged, or have to be brought home (& they DO have another
one) prematurely.
It's in the subject line
Layered classification nets have always struck me as the right approach, particularly as we learn more about how human senses work - it seems like a lot of our "thinking" is done much closer to our sense organce than we might have once imagined. Interesting that the less "organic" type, decision trees, were used rather than neural nets. One wonders if maybe it was more a matter of ease of phrasing/training/debugging than of classification itself that decided which type to use.
I have more than 1000 cores in my desktop lol maybe they just hooked their desktops together
Smells like Neural Networks thinking ...
- "What do you do for a living?"
- "I train trees to make a decision forest that can see human limbs."
- "Ah, I see. Makes sense. (WHAT THE FUCK???)"
I want to shit on your chest as I cum in your mouth.
From the summary it looks like they are basically using a classifier which they spent a lot of time training, and it works well. This is impressive, but I don't know if it meets the story title's claim of "AI breakthrough", since from the summary it sounds basically like, "researchers used classifier for classifying data and it worked!" Can someone summarize in a little more detail exactly what the "breakthrough" entails, other than basically standard use of classifiers for training on data sets?
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.
Ummm, all I've seen so far apart from this are pretty obvious uses of the depth sensor.
What Microsoft has done is solved an extremely hard AI problem. Check out the body-part identification. I think more credit is due.
Trees have traditionally been trained in Entish, which although reliable, is such an un-hasty language.
Modest doubt is called the beacon of the wise. - William Shakespeare
A lot of the MS-haters on Slashdot tried to write off the Kinect as a nice bit of third-party hardware with a crappy MS-made driver. I wonder how they'll respond to this. Microsoft has really outdone themselves here. I think Penny Arcade put it best. If only they could apply this sort of innovation to their more important products, they'd be back on top in no time.
I haven't thoroughly read the paper yet, but calling this an AI breakthrough is inappropriate for a number of reasons. First, this is an application of machine learning, which is not the same thing as AI. Second, it seems to be a fairly incremental work building on very common techniques--very far from a breakthrough in any respect. If you don't believe me, see some of Jamie Shotton's other work, which is good work, but this is nothing extraordinary in comparison.
I thought everything was developed by a third party? So why is MS getting all the credit? Not to mention this doesn't sound like a breakthrough since it is using some common, and old, techniques.
IP (#35625124) resolved to slashdot user Wovel (964431).
So...it can't see the forest for the limbs?
Neural Network / perceptrons.
I would like to know why they choose a relatively slow method, random forests (RF), over something like an SVM or another classifier based on a convex optimization? They claim the RF run fast on the GPUs, although they also mention reference [6], which uses SVMs on this problem. Are the RF actually faster than a modern, non-linear SVM implementation?
And is this how they got the images tagged in the first place? http://kotaku.com/#!5605936/is-this-how-microsoft-will-fix-kinects-couch-problem
"[..] the decision trees were modified until they gave the correct classification for a particular body part across the test set of images"
this is called cheating in machine learning (you are not allowed to modify your model(s) based on the results on the test set).
and of course it is not what they do.
nice piece work, tho IMHO not AI breakthrough.
Being possessed of an enormously large penis, I am unable to use Kinect as it keeps detecting it as a third leg!
Gentoo Linux - another day, another USE flag.
That 1000 core system used to train Kinect... what version of Windows was running?
"I think this line is mostly filler"
The method they are using s called as haar cascades postulated by viola jones. I have used the same with opencv for a bit now. http://en.wikipedia.org/wiki/Haar-like_features It's basically passing An image thru progressive classifiers to get a final weight of match. Microsoft may have done the training for generating the classifiers but the method has been around for a bit. "Decision tree".... Pfffft.
Why can't MS do stuff like this in all their departments? Are there not enough smart people to go around? You get truly cool things like this, juxtaposed with
lame "us too!" attempts like WP7 and Bing.
Before you design for reuse, make sure to design it for use.
So which side does it use, left or right?
Way more hyperbole was used to describe the feats of the Wii remote than in the Penny Arcade comic, even before Motion Plus was released...