3 ms·
They're likely using a commercial facial recognition system which includes that feature for demos at trade shows. People love getting external validation, which
by yorwba 8y ago
They're likely using a commercial facial recognition system which includes that feature for demos at trade shows. People love getting external validation, which also explains why the scores are all greater than 70.
Edit: E.g. this article mentions a system being demoed with that feature: https://www.npr.org/sections/parallels/2018/04/03/598012923/facial-recognition-in-china-is-big-business-as-local-governments-boost-surveilla https://www.npr.org/sections/parallels/2018/04/03/598012923/...
- saagarjha 8y agoNumbers don’t really say much unless you pin them to what they’re measuring: they might not be calibrated against a percentile. For example, on a linear scale with 0 being “horrendous” and 100 being “insanely attractive”, I’d rate most of the people I know much closer to 70 than a 50: the average person could stand to look a lot worse.
- bluntfang 8y ago>Numbers don’t really say much unless you pin them to what they’re measuring That's why it has the label "attractive"
- simonh 8y agoSigh, there's always one. The point is that's not a sufficiently clear definition of what's being measured. Someone I find attractive you might find unattractive, because it's a subjective attribute, so 'attractiveness' isn't an unambiguous description of what the algorithm is actually measuring.
- johncolanduoni 8y agoIf it’s a feature for making people who try your system at a trade show feel good about themselves as they said, it works best if it’s not measuring anything at all.
- bluntfang 8y agohave you read the thread you're participating in? I digress, surely naming is really hard. there's a sweet spot to hit short and sufficiently explanative. Perhaps this didn't hit it for you, but it probably did for the target audience, and if it didn't, they could call support for the product they paid for.