How much do you think McDonalds would pay for a magic camera + menu that boosted all their store revenue by 5% overnight?
Sorry, but I'm just not buying the vision. Facebook has ridden this incredible wave, first of investors spooging themselves with excitement over the potential value of all the data collected, then the wave of advertisers coming in and spending massive amounts with very questionable ROI. Long term everyone is going to get more sophisticated and realize there are cases where "would you like fries with that?" simply performs better once you factor in the weaknesses of comprehensive data-driven in-store advertising (cost, bandwidth, alienation, noisy data, key data out of band, etc, etc).
I'm not suggesting it's a good idea, just that it's an option.