Learning OpenCV 4 Computer Vision with Python 3 - Third Edition

Using CamShift

Although CamShift is a more complex algorithm than MeanShift, OpenCV provides a very similar interface for the two algorithms. The main difference is that a call to cv2.CamShift returns a rectangle with a particular rotation that follows the rotation of the object being tracked. With just a few modifications to the preceding MeanShift example, we can instead use CamShift and draw a rotated tracking rectangle. All of the necessary changes are highlighted in bold in the following excerpt:

import cv2import numpy as np# ... Initialize the tracking window and histogram as previously ...success, frame = cap.read()while success:    # Perform back-projection of the HSV histogram onto the frame. hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV) ...