A big biometric security company in the UK, Facewatch, is in hot water after their facial recognition system caused a major snafu - the system wrongly identified a 19-year-old girl as a shoplifter.
A big biometric security company in the UK, Facewatch, is in hot water after their facial recognition system caused a major snafu - the system wrongly identified a 19-year-old girl as a shoplifter.
This makes it sound like you only tried one particular set of twins–unless there were multiple sets, and in each set the two had very different styles? I’m no statistician, but a single set doesn’t seem statistically significant.
What I’m saying is we had a deployment in a large facility. It was a partnership with the org that owned the facility to allow us to use their location as a real-world testing area. We’re talking about multiple buildings, multiple locations, and thousands of people (all aware of the system being used).
Two of the employees were twins. It wasn’t planned, but it did give us a chance to see if twins were a weak point.
That’s all I’m saying. It’s mostly anecdotal, as I can’t share details or numbers.
No, it gave you a chance to see if that particular set of twins was a weak point.
With that logic we would need to test the system on every living person to see where it fails.
The system had been tested ad nauseum in a variety of scenarios (including with twins and every other combination you can think of, and many you can’t). In this particular situation, a real-world test in a large facility with many hundreds of cameras everywhere, there happened to be twins.
It’s a strong data point regardless of your opinion. If it was the only one then you’d have a point. But like I said, it was an anecdotal example.