OpenCV 4 with Python Blueprints - Second Edition

Implementing a Kalman filter

Now, geared with our model, let's get our hands dirty and write a class that handles all this magic. We are going to write a custom class that will use cv2.KalmanFilter as a Kalman filter, but we will add some helper attributes to be able to keep track of each object.

First, let's take a look at the initialization of the class, where we will set up our Kalman filter by passing the state model, transition matrix, and initial parameters:

1. We first start by initializing the class with the boundary box—bbox—and the label for the label object:

class KalmanBoxTracker:    def __init__(self, bbox, label):

2. Then we set up some helper variables that will let us filter boxes as they appear and disappear in the tracker: