How to use trained SVM classifier with vision.ObjectDetector?

2 views (last 30 days)
I want to train a SVM classifier (say using classifier = fitcecoc(trainingFeatures, trainingLabels);). Feature type is not currently important, say HoG. Then I want to use this classifier for object detection. I don't want to use the system vision.CascadeObjectDetector as this is a decision tree based detector that doesn't return confidence values per object proposal. Is there an alternative that I have missed? Any proposals will be greatly appreciated.

Accepted Answer

Dima Lisin
Dima Lisin on 18 Nov 2015
You can certainly train your own HOG-SVM classifier. The bad news is that to use it for object detection, you would have to write your own code to do the sliding windows.
  2 Comments
AD
AD on 18 Nov 2015
That's what I thought. Thanks for the answer. It is a shame though, as multiscale sliding window search is obviously already efficiently implemented in vision.ObjectDetector. Why not expose it?

Sign in to comment.

More Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!