For the threshold value, how is 0.25 determined to be the most optimal? is this value just specific to the datasets YOLO was trained on or is it something that is universal.

Also, this is really awesome!