OpenCV and Python : Hog Compute error 215

Question

I'm trying to use OpenCV SVM with Hog descriptors. Working fine on a first sample with Python 2.7. When trying on an other example on python 3.6.1, i'm getting the following error :

 error: /usr/local/src/opencv/modules/objdetect/src/hog.cpp:241: error: (-215) img.type() == CV_8U || img.type() == CV_8UC3 in function computeGradient 

Here is a bunch of the code :

print('Defining HoG parameters ...')
print (np.array(digits_deskewed).shape)
print(digits_deskewed[0])
# HoG feature descriptor
hog = get_hog();
hog.compute(np.array(digits_deskewed[0]))

Gives this output :

(42000, 28, 28) [[ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.] [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.] [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.] [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.] [ 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 188. 255. 94. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]

With get_hog() as :

def get_hog() :
    winSize = (28,28)
    blockSize = (4,4)
    blockStride = (4,4)
    cellSize = (8,8)
    nbins = 9
    derivAperture = 1
    winSigma = -1.
    histogramNormType = 0
    L2HysThreshold = 0.2
    gammaCorrection = 1
    nlevels = 64
    signedGradient = True

    hog = cv2.HOGDescriptor(winSize,blockSize,blockStride,cellSize,nbins,derivAperture,winSigma,histogramNormType,L2HysThreshold,gammaCorrection,nlevels, signedGradient)

    return hog

I have no idea of the reason of this error, can someone give me a clue ?

Thanks !

Answer 1

you have float type in your numpy array. you need uint8.