All missing the actual technology point here. From a technology perspective this isn't about whether arrests took place, someone covered their face or any of that, it's simply put, did the technology correctly match the face to the data already in its database? This technology trial would be much better served at a large institution with hundreds/thousands of participants who have willingly agreed to take part, like a university or national/local government building. Set the platform up with a count of 10% of the total population and let it do it's thing. Facial recognition is simply that, what happens with the result after that is down to the individual user (police making arrests, casinos ejecting known card counters etc.).