How to use EKF for SORT Algo

[iamrajee]

Hello Everybody, I'm using SORT for tracking the bounding boxes detected by YOLO. (code)

The previous implementation involves using KalmanFilter from filterpy, However I would want to replace that with ExtendedKalmanFilter.

Here is my new code, Where my dimension are as:

self.kf = ExtendedKalmanFilter(dim_x=7, dim_z=4)
    self.kf.F = np.array([[1,0,0,0,1,0,0],[0,1,0,0,0,1,0],[0,0,1,0,0,0,1],[0,0,0,1,0,0,0],  [0,0,0,0,1,0,0],[0,0,0,0,0,1,0],[0,0,0,0,0,0,1]])
    self.kf.H = np.array([[1,0,0,0,0,0,0],[0,1,0,0,0,0,0],[0,0,1,0,0,0,0],[0,0,0,1,0,0,0]])

    self.kf.R[2:,2:] *= 10.
    self.kf.P[4:,4:] *= 1000. #give high uncertainty to the unobservable initial velocities
    self.kf.P *= 10.
    self.kf.Q[-1,-1] *= 0.01
    self.kf.Q[4:,4:] *= 0.01

    self.kf.x[:4] = convert_bbox_to_z(bbox)self.kf = ExtendedKalmanFilter(dim_x=7, dim_z=4)
    self.kf.F = np.array([[1,0,0,0,1,0,0],[0,1,0,0,0,1,0],[0,0,1,0,0,0,1],[0,0,0,1,0,0,0],  [0,0,0,0,1,0,0],[0,0,0,0,0,1,0],[0,0,0,0,0,0,1]])
    self.kf.H = np.array([[1,0,0,0,0,0,0],[0,1,0,0,0,0,0],[0,0,1,0,0,0,0],[0,0,0,1,0,0,0]])

    self.kf.R[2:,2:] *= 10.
    self.kf.P[4:,4:] *= 1000. #give high uncertainty to the unobservable initial velocities
    self.kf.P *= 10.
    self.kf.Q[-1,-1] *= 0.01
    self.kf.Q[4:,4:] *= 0.01

    self.kf.x[:4] = convert_bbox_to_z(bbox)

And calling the function as

self.kf.update(convert_bbox_to_z(bbox), HJacobian=self.H_of, Hx=self.Hx)

Where I don't really don't know how to write H_of and Hx functions. Thus it would be great if you could help me write these two functions. My current dummy definition is as below:

def H_of(x,y):
     H = np.array([[1,0,0,0,0,0,0],
                [0,1,0,0,0,0,0],
                [0,0,1,0,0,0,0],
                [0,0,0,1,0,0,0]])
    return H
def Hx(x,y):
    Hx = np.array([[0],
                [0],
                [0],
                [0]])
    return Hx

Thanks!