AbbeFaria's profile - activity

Interesting, normally when mine are in that range the cascade picks up a lot of false positives, or sometimes nothing at all. How large are your sample sets generally?

I've been working with Object Training for a while now, and I've found that generally cascades with an acceptance ratio of .0002 to .0007 are the most refined generally. However, I don't quite understand how acceptance ratio is calculated using traincascade. Could someone explain it to me or point me to a good reference that explains it?

I've been working with Object Training for a while now, and I've found that generally cascades with an acceptance ratio of .0002 to .0007 are the most refined generally. However, I don't quite understand how acceptance ratio is calculated using traincascade. Could someone explain it to me or point me to a good reference that explains it?

Hey, here's my solution to convert between the two:

List<MatOfPoints> data = new ArrayList<MatOfPoints>();
List<MatOfPoint2f> matrix = new ArrayList<MatOfPoint2f>();

Imgproc.findCountours(src,data,new Mat(),Imgproc.RETR_LIST,Imgproc.CHAIN_APPROX_SIMPLE);

for(int i = 0; i < data.size(); i++){
    MatOfPoint2f myPt = new MatOfPoint2f();
    data.get(i).convertTo(myPt, CvType.CV_32FC2);
    matrix.add(myPt);
}

I dont think so, your going from List to just a MatOfPoint2f.

What kind of applications would one use calcCovarMatrix() for? I did some research and I mostly understand how Co-variance Matrices work, but couldn't find much information on applications of it or things this type of calculation could be used for.

You should look up each of the feature types and do some research to understand how they work,and then ask more detailed questions. I'll put some links here for you. There are so many different variables about your embedded system that are unknown to us that it makes it impossible to gauge what will be the best. It will take time and experimentation, really, no matter what. Depending firstly on your system constraints and what kind of features you wish to detect. HoG: http://en.wikipedia.org/wiki/Histogra... HAAR: http://en.wikipedia.org/wiki/Haar-lik... LBP: http://en.wikipedia.org/wiki/Local_bi... Good Luck!

Ah, to follow up. I ended up with a very good LBP Cascade that works with stunning accuracy. Thanks for that.

Steven, thanks for such a good explanation. Have a similar issue over here, however the solution was a drastic decrease in my numPos sample size. I went from 5272 to 1500. Anything higher than that and it would kick back this error. I thought you would find this interesting. I was wondering your opinion on this. I'm training to detect american license plates at the moment. It's one of the first objects I'm training for. Do you think that images from this link [http://www.plateshack.com/platelist.html] (just American Car Plates) are suitable for training? Or does lacking different background cause issues? Perhaps, could the features on the plates themselves be so different that the software rejects most of my data while creating samples? Cause that's what I'm thinking. Training for LBP.