Comments on: “Computer vision not as good as thought”, who thought?! http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/ Computer vision, image analysis, and related mathematics Thu, 17 Feb 2011 13:25:51 +0000 hourly 1 http://wordpress.org/?v=3.1.3 By: Computer Vision for Dummies » Why machine learning never works http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/comment-page-1/#comment-74 Computer Vision for Dummies » Why machine learning never works Fri, 15 Feb 2008 20:22:49 +0000 http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/#comment-74 [...] In response to my previous post Bob Mottram wrote “Using many learning algorithms (genetic, neural, etc) it is very easy to categorise images on some very trivial basis. In theory the larger the data set the harder the system has to work and the less likely it is to find a quick “cheat”, but it all depends upon how features are being represented in the system.” [...] [...] In response to my previous post Bob Mottram wrote “Using many learning algorithms (genetic, neural, etc) it is very easy to categorise images on some very trivial basis. In theory the larger the data set the harder the system has to work and the less likely it is to find a quick “cheat”, but it all depends upon how features are being represented in the system.” [...]

]]> By: Peter http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/comment-page-1/#comment-71 Peter Fri, 01 Feb 2008 13:38:07 +0000 http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/#comment-71 I meant "machine learning". I meant “machine learning”.

]]> By: Peter http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/comment-page-1/#comment-70 Peter Fri, 01 Feb 2008 04:11:59 +0000 http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/#comment-70 Thanks for the feedback, Bob. I think a reply may warrant a separate post. I will use this opportunity to clarify my thinking on machine vision a bit. Thanks for the feedback, Bob. I think a reply may warrant a separate post. I will use this opportunity to clarify my thinking on machine vision a bit.

]]> By: Bob Mottram http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/comment-page-1/#comment-69 Bob Mottram Thu, 31 Jan 2008 22:46:06 +0000 http://inperc.com/blog2/2008/01/26/%e2%80%9ccomputer-vision-not-as-good-as-thought%e2%80%9d-who-thought/#comment-69 Using many learning algorithms (genetic, neural, etc) it is very easy to categorise images on some very trivial basis. In theory the larger the data set the harder the system has to work and the less likely it is to find a quick "cheat", but it all depends upon how features are being represented in the system. To achieve invariance across many images what the system needs to do is hypothesise the local surface normal of each observed feature and build those into a 3D geometric hash. Using many learning algorithms (genetic, neural, etc) it is very easy to categorise images on some very trivial basis. In theory the larger the data set the harder the system has to work and the less likely it is to find a quick “cheat”, but it all depends upon how features are being represented in the system.

To achieve invariance across many images what the system needs to do is hypothesise the local surface normal of each observed feature and build those into a 3D geometric hash.

]]>