s****i 发帖数: 150 | 1 我这也是网上搜到的,不是内部资料。非常奇怪居然很多人看不起kinect。。。以后任
何一本machine learning或者computer vision或者graphics的书,肯定都会提到它的。
3个关键:1) pixel be processed independently on GPU, 2) work with all ages/
body shape/clothing, 3) recognition does not rely on any temp infor.
做ML的都知道,很多时候所谓的high recall/precision算法都是靠选择性挑选数据做
出来的,而且往往不能重复;kinect这个绝对不是靠照着几篇paper做做就可以搞得定
的。而且程序本身的效率很得很高。。。
下面是原文。链接在这里http://www.inf.ethz.ch/news/colloquium/details/index_DE?id=1246。Jamie Shotton本身是个大牛,他的paper list在这里:http://research.microsoft.com/en-us/people/jamiesho/publications.aspx
Deriving from our earlier work that uses machine learning to recognize
categories of objects in photographs, body part recognition uses a
classifier to produce an interpretation of pixels coming from the Kinect
depth-sensing camera into different parts of the body: head, left hand,
right knee, etc. Estimating this pixel-wise classification is extremely
efficient, as each pixel can be processed independently on the GPU. The
classifications can then be pooled across pixels to produce hypotheses of 3D
body joint positions for use by a skeletal tracking algorithm. Our approach
has been designed to be robust, in two ways in particular. Firstly, we
train the system with a vast and highly varied training set of synthetic
images to ensure the system works for all ages, body shapes & sizes,
clothing and hair styles. Secondly, the recognition does not rely on any
temporal information, and this ensures that the system can initialize from
arbitrary poses and prevents catastrophic loss of track, enabling extended
gameplay for the first time. | t*********l 发帖数: 566 | 2 这个要顶
哥做vision做了四五年。看到kinect的demo就知道它的contribution有多大。
深度图人人都能做,能做到这个家庭娱乐程度工业级别的,没有别家。 | P***P 发帖数: 1387 | 3 小声说一下, rgbd做深度图研究的, msr还真比不上intel research | W***i 发帖数: 9134 | |
|