Machine Vision made Easy - SimpleCV
simplecv.org
simplecv.org
1) We will probably release the SimpleCV 1.2 super packs next week. There are a ton of sweet new features. We could really use some beta testers. Please let us know if you are interested.
2) We just signed a book deal with O'Reilly Media to make a SimpleCV guide. We are soliciting input from the community about what CV problems they need solved, or cool projects they would like us to do.
With respect to our business model, right now we are focused on quickly and cheaply delivering open-source machine vision solutions for manufacturing. If you have a manufacturing contact that has a machine vision problem and is looking for a solution we would love to talk to them. In the next month we will have a development scrum that will focus on our Seer product. Seer will the framework for deploying your computer vision projects in a cloud context. This means being able to remotely deploy and manage a vision system (i.e. CV as a service). Also along these lines we are actively looking for partners for the DARPA iFAB BAA that will be released soon (see: https://www.fbo.gov/index?s=opportunity&mode=form&id...). If you are going to submit a proposal for this contract we would love to talk to you.
Since we're open source we welcome and encourage community participation and feedback. If you have a cool project or a pony request shoot us a message via twitter (@Simple_CV or @IngenuitasInc) and we can chat and perhaps help you with your project.
http://hplusmagazine.com/2011/09/20/make-computers-see-with-... (4th comment)
/me might be in the recruiting realm too long
There's no way I could get used to CV not being 'curriculum vitae'.
Examples include Intel's OpenCV and the CVPR conference.
(I actually thought that it was weird that they called themselves "machine vision" rather than computer vision in the title.)
Machine vision usually refers to the domain of industrial inspection where the scene contents are highly controlled, and computer vision usually refers to harder problems where someone is walking around a real environment with bad lighting and awkward perspectives.
Some researchers prefer the term 'machine vision' as more general, referring to the entire field of artificial vision problems, not necessarily involving a 'computer' in common terms (i.e. FPGA-based or analog electronics).
https://github.com/ingenuitas/SimpleCV/blob/master/SimpleCV/...
> def _makeCodebook(self,data,ncodes=128):
> """
> Do the k-means ... this is slow as as shit
> """
It certainly looks easier than starting with OpenCV alone.
Moreover, Ingenuitas looks like an interesting company. Open source solutions for manufacturing is certainly nothing I've heard of before. Definitely something to keep an eye on.
I also wanted to mention that the team (Kat, Anthony et al.) are very responsive to their community and user base.
All around kudos SimpleCV team.
SR
in my experiecne the hardest part of machine vision (on a pC) is getting all the bits installed and playing nicely with each other
they don't teach you that the defualt opencv built with vs will clash with Qt libs built with mingw but the ing system won't tell you this!
Of course back in the day, when it had to be a Sparc station crammed full of custom i860 boards it as even harder!
SimpleCV just released a video about face detection but I don't think they've expanded into features yet. Hopefully they don't cause I want to do it!
The extracted features appear to be nothing but small patches of various types and directions of edges.
Training a Naive Bayes classifier on these features will try to determine the label roughly based on the unordered strength/number of occurences of these small feature patches.
While it's obviously demonstrated that you can attain a surprising accuracy with this method, if the features in the "bag of features" method are all very local like in this demo, you are throwing away the larger scale features of the image. And intuitively, it's the larger scale features that ultimately determine the subject of a photograph.
One thing this classifier would probably fail, would be a blurred photograph, that still would obviously depict either a cat or a cheeseburger to a human viewer, but it would completely mess up the local feature patches.
A more interesting, and much harder, thing would have been to try and train a classifier to determine "cat", "cheeseburger" or "other/neither". I'm fairly sure you couldn't have done that with such local features.
Just my educated Machine Learning opinion, of course :-) The demo itself is both hilarious, and a good demonstration of SimpleCV's ease of use.