For me the easiest way is to take all the values returned by image.shape:
height, width, channels = img.shape
if you don't want the number of channels (useful to determine if the image is bgr or grayscale) just drop the value:
height, width, _ = img.shape