You're assuming the primary goal of the intelligence is to identify the suspects. Sure, that would be nice, but a more likely outcome is to track the suspects to their next stop, and then start again with the video analysis to see who got out of the car and where they went next...
Biting the hand that feeds IT
