e (sequence of floats, optional): tuple of two floats defines minimum and maximum factors to
            scale the input size.
        p (float, optional): probability of the input being flipped. Default value is 0.5
    r