There is another method, This used ultrasonic transducers in close contact with the bottom of the bottle. Just like a submarine, a transducer sends out a high frequency pulse, which traverses the thickness of the bottom wall of the bottle and into the liquefied gas. At every surface the sound wave meets, the change in density of the substance causes some of the sound wave to be reflected back to the transducer - so to get one from the inside wall of the gas bottle, the top of the liquefied gas and the top of the vapour- the top of the bottle.
By measuring the time it takes for the waves to be reflected back to the transducer, the various surfaces can be accounted for. The metal of the bottle has a very short time interval, the top of the liquefied gas is quite a lot slower, and the inner surface of the top of the bottle will be significantly weaker and slower.
By filtering out the short time interval signal and ignoring the low energy reflections from the top and sides of the bottle, the remaining signal is the reflection from the top of the liquefied gas. It is not too difficult with modern electronics to turn that into a usable indication of how full or more accurately how close to empty the bottle is.
I believe that Truma market a product that uses a similar principal.