Currently size of image is resized with cv2.resize, but it's using constants, instead of camera input dimensions (cv2.VideoCapture(0)) to scale from