joe minichino's profile - activity

NAME cv2.bgsegm

FILE
    (built-in)

FUNCTIONS
    createBackgroundSubtractorGMG(...)
        createBackgroundSubtractorGMG([, initializationFrames[, decisionThreshold]]) -> retval

    createBackgroundSubtractorMOG(...)
        createBackgroundSubtractorMOG([, history[, nmixtures[, backgroundRatio[, noiseSigma]]]]) -> retval

looking at

detectMultiScale(...)
    detectMultiScale(image[, scaleFactor[, minNeighbors[, flags[, minSize[, maxSize]]]]]) -> objects

it seems like you're just getting faces (and nothing else) back, and there's no outputRejectLevels argument.

I'm operating object detection on an image using svm and sliding windows (with opencv 3 and python)

When testing a region of an image with svm predict i get a classification and a score (i.e. Class: 1, Score: -1.035665 ), which I obtain with svm.predict( features, flags=cv2.ml.STAT_MODEL_RAW_OUTPUT ) I want to apply non-maximum suppression on overlapping regions, but I'm a bit fuzzy about how to rate the highest score, seeing as higher values don't really correspond to better accuracy, and it seems like the most accurate windows are around -0.5 score (but this is a conclusion I drew empirically). Documentation on this isn't abundant, any clarification would be awesome.

@berak thanks again, I was getting a bit adventurous there. As i said in the question i had a hunch I was misunderstanding the process.

thanks @berak - is there a particular way to serialize the svm?

I successfully trained a svm with BOW in python (opencv 3). Now I would like to pass it to a hog with setSVMDetector. Is this possible or am I misunderstanding the use of the method? when i try to pass a trained svm to a hog detector like this:

svm = cv2.ml.SVM_create() # opencv 3.0-rc1 API
"""train svm..."""
hog = cv2.HOGDescriptor()
hog.setSVMDetector(svm)

i get TypeError: _svmdetector is not a numpy array, neither a scalar

I tried saving the svm to xml and then reload it and discovered the load() method in opencv 3 and discovered the load() method of SVM is gone.

thanks @berak, updating opencv worked. For the sake of accuracy to anyone who may wonder about the same thing, you now create an svm in python with cv2.ml.SVM_create()

thanks a lot for the quick response @berak (as usual). I am currently building it, for some reason i was getting the same error as here so I cleaned up the build folder and started from the beginning. Will revert when it's finished.

Anybody knows where cv2.SVM() is gone? if I do a >>> help(cv2) it's not there. I would think it may be included in ml but ml only contains enum values.