Key points from the video:
- It is primarily a promotion for UK manufacturer Digital Barriers, which is surely great press for them.
- They show facial surveillance but it is a super simple demo setup which shows nothing about how it would really work in production, at scale.
- The impetus of the segment is the recent terrorist events in the UK but they admit that the terrorists were "on the radar but not under the microscope when they attacked", so even if they used facial surveillance and it worked, it not clear what they would have done with them since they had not yet committed nor were wanted for committing a crime.
- Towards the ends, they suggest that facial surveillance could also be used for search / intelligence gathering to "determine patterns of how often people visited a location" though the likelihood that the system missed potential suspects and undercounts activities is high.
- It ends with the obligatory privacy rights advocate but no discussion at all about how well the system works at scale. The demos shows matches at .51 or .65 confidence, which in a real deployment scanning tens of thousands of faces daily and with numerous people on watch lists raises questions about the level of false alerts.
Any thoughts or questions on this?