As stated by yourself, the coordinates of the bounding box of any returned face/eye detection is of the form x_upper_left_corner y_upper_left_corner width height keeping in mind that the origin of OpenCV is at the upper left image corner.