Yes that is how gain is usually defined for camera sensors It is slightly perverse because the figure goes down as the conventional gain (how much the signal is amplified) increases !
With CCD sensors the gain is normally fixed, set so the full well depth of the sensor (the point at which the sensor becomes saturated) matches the maximum counts of the ADU.(In this case it would at 5e-/ADU for a 12 bit ADU and FWD of 20k e-) But with CMOS sensors you can increase it which reduces the maximum number of electrons you can count before you reach the maximum of the ADU. Why would you want to do that you may well ask? Well with CMOS the read noise goes down as you increase the gain so in circumstances where you cannot use the full well depth (eg a faint object or you need a short exposure) you can increase the gain and take advantage of the lower read noise.
Cheers
Robin
