pts = np.float32([[0,0],[0,h],[w,h],[w,0]]).reshape(-1,1,2)
pts_new = cv2.perspectiveTransform(pts, H)
What is the most efficient way
