calculate body part position using temporal data

[MasterMedo]

Instead of trusting the MoveNet blindly, we can make the position of each body part a weighted exponential moving average that depends on the last position of the body part, current position of the body part and the confidence. The weight of the exponential equation will depend on the confidence of the detected body part. For example, if the confidence is above 60%, then the weight should be big enough that we consider the body part basically at that exact position. Otherwise, we move the body part from the last position towards the new position based on the confidence.

Also, it would be great to limit the speed a body part can move at! This can be in cm/s!

Create a function:

def adjust_body_part_positions(body_parts: np.array) -> np.array:
    """Calculates the body part positions based on temporal data.
    It takes into account:
        - historical position(s) of the body parts
        - current position of the body parts
        - confidence levels
        - maximal speed a body_part can move at
    """
    global body_parts_history