3 ms·
Seems similar to the 2017 test in Berlin https://www.dw.com/en/facial-recognition-surveillance-test-extended-at-berlin-train-station/a-41813861 https://www.dw.c
by mtmail 4y ago
Seems similar to the 2017 test in Berlin https://www.dw.com/en/facial-recognition-surveillance-test-extended-at-berlin-train-station/a-41813861 https://www.dw.com/en/facial-recognition-surveillance-test-e...
From the test strategy document "Cohort subjects will be drawn from two sources (i) an actor’s agency and (ii) an under 18 cohort of volunteers from the Police Cadets. 400 Cohort Subjects will be needed to achieve the Trial Objectives."
In Berlin it was 300 and people could volunteer.
If I remember the Berlin setup correctly the train station was used to have a high number of random, non repeat people, as the goal is to recognize people from a small subset
The document doesn't list the location (Westminster) so where the test is done must have been a separate decision.
- hansword 4y agoThe berlin sudkreuz test was a disaster, by the way. * Moving your head more than 15deg away from the camera was enough to slip through. * false positive rate was 0.67% - you can calculate yourself how that works on one of the largest train stations in Berlin serving close to 100k people per day (German language source, can't find en translation) https://www.ccc.de/en/updates/2018/debakel-am-suedkreuz https://www.ccc.de/en/updates/2018/debakel-am-suedkreuz
- DocTomoe 4y ago> * Moving your head more than 15deg away from the camera was enough to slip through. Which can be solved by adding more cameras so there's no angle in which your face is not in full view.