General Cascade Classifier for any type of features
I think it would be important for OpenCV to add a boosted cascade classifier to its functionalities. I mean a general algorithm, and not one that only works with haar features.
Does anyone know if something like this is going to be implemented any time soon?
EDIT: is it possible to use any type of features on the CascadeClassifier method? If so, is there any tutorial explaining how to do so?
I've created an algorithm that extracts a very large pool of features from various channels of the images. The features resemble Haar Wavelets but differ in the fact that in HW are extracted only from intensity images, whereas my features are extracted from 10 different channels, so you can see the difference.
I've used the boost class to create a predict method that works nicely in terms of detection rate, but it is very very slow because I always have to evaluate the whole set of features. Therefore I wanted to train a boosted cascade classifier in order to speed up the process. But almost all references I find online about cascades and openCV are for Haar Wavelets, mainly the Viola and Jones implementation for face detection.4
I have my data in a .csv file with ~15K classified samples and 15K features each. Is it possible to use this file to create a cascade classifier?
EDIT 2: I cant use detectMultiScale function because I need to evaluate my own algorithm.
