2

I am trying to train an adaboost classifier using the openCV library, for visual pedestrian detection. I've come across the notion that adaboost allows the selection of the most relevant features, meaning, if I harvest 50.000 features from images and then use them to train a classifier, in the end of the training process I would be able to select, for example, the best 2000 out of those 50.000.

Then, this would allow me to harvest only those 2000 during the actual process for the sake of speed.

Is this even true? Or am I falling in a misconception?

If true,, is it possible to be done using the openCV library?

Best regards

Pedro Batista
  • 1,100
  • 1
  • 13
  • 25

1 Answers1

4

Yes, this is true. That's exactly what boosting is all about.

Please, check the OpenCV documentation about training a cascade of boosted classifiers.

Ramiro
  • 698
  • 6
  • 21
  • I am not understanding that reference manual,I will soon look at some tutorials. But imagine I have a .txt file with some thousand of classified samples and each sample a great number of features. Would this method you refer work with my data? Also, it seems to me that this cascade generator is prepared to work with Haar wavelets, but my features are kinda different. Haar Wavelets are computed only in grayscale images, and my features are computed over different channels of the image. – Pedro Batista May 08 '13 at 03:55
  • I'm not sure if OpenCV is ready to read your data as it is. Also, I'm not sure if OpenCV implementation is able to work with any sort of feature. AFAIK, you can implement your own feature and plug it in OpenCV, but I never did that. Anyway, boosting - as a mathematical concept - does what you want: from a large set of features, it selects those that minimize the classification error, as long as you have enough data to train it. – Ramiro May 19 '13 at 20:13