how to pass keypoints for orb.computeAsync

asked 2020-05-04 03:45:23 -0600

updated 2020-05-04 03:47:17 -0600

Im trying to compute descriptors for the keypoints found, but Im getting segmentation fault error. Part of the problem is that there is very little documentation for opencv cuda in python. I am getting an image data, and using cv2.cuda.createGoodFeaturesToTrack() for getting keypoints. I want to calculate descriptors for the keypoints, and hence using computeAsync(). Here is my code -

def extract(img):
    #orb = cv2.cuda_ORB.create()
    orb = cv2.ORB_create()
    #converting to grey
    cuC1 = cv2.cuda_GpuMat()
    cuC1.upload(img)
    cuC1 = cv2.cuda.cvtColor(cuC1, cv2.COLOR_RGB2GRAY)
    #detection
    pts = cv2.cuda.createGoodFeaturesToTrackDetector(cv2.CV_8UC1)
    _kps = pts.detect(cuC1)
    npKps = _kps.download()
    npKps = np.int0(npKps)
    for n in npKps:
        #extraction
        kps = [cv2.KeyPoint(x=f.ravel()[0], y=f.ravel()[1], _size=20) for f in n]  #cpu keypoints, works well with non cuda orb
    kps, des = orb.computeAsync(cuC1, _kps)
    #return points and descriptors
edit retag flag offensive close merge delete