Blog Post #411:  How good is facial recognition getting?

Blog Post #411:  How good is facial recognition getting?

Taken from:  https://www.nytimes.com/2018/02/09/technology/facial-recognition-race-artificial-intelligence.html

Facial recognition is getting good.  

Let’s think about that for a second.  In the past, fingerprints were the common ‘perfect’ recognition method.  (And, that is probably still the case). Fingerprints are unique. But, crooks know that and use gloves, wipe down surfaces where they might have left a fingerprint.

We have other ‘maybe’ unique identifiers – like legal names (but, I’m friends with two other people named “Bruce White” on Facebook – even with my middle name, that isn’t quite unique.  I just checked, and in South Korea, the most common last names are Kim, Lee, and Park – comprising over 50% of all surnames in South Korea.

Well – how about the uniqueness of Social Security numbers?  My research found this statement: “Today, assigned randomly and never recycled, a social security number is as unique an identifier as your fingerprints.”  (There is a small caveat – in the past, there were a few non-unique numbers by accident). But, another article suggested that for valid reasons there have been people who have been given a second social security number (like witness protection programs).

The New York Times article states:
“Facial recognition technology is improving by leaps and bounds. Some commercial software can now tell the gender of a person in a photograph.  When the person in the photo is a white man, the software is right 99 percent of the time.”

But that article also suggests that the accuracy declines with non-white faces:
“But the darker the skin, the more errors arise — up to nearly 35 percent for images of darker skinned women, according to a new study that breaks fresh ground by measuring how the technology works on people of different races and gender.”  Later in the article, it says “In 2015, for example, Google had to apologize after its image-recognition photo app initially labeled African Americans as gorillas.”

The huge advantage of facial recognition systems is that you don’t have to have physical records (like fingerprints).  With the speed of computers, you can scan and identify a group of people in Grand Central Station in New York City quickly and unknown to the people.   JetBlue airlines has experimented with using Facial Recognition to board planes – rather than boarding passes (or even electronic boarding passes).

Gate agent to people boarding the flight:  “Look into the camera”; Passengers look into the camera – which identifies them as a passenger.  Gate agent (again) says “You are in seat 24C” (also displayed on a screen). No digging out your cell phone for your boarding pass, no digging through your pockets, purse or backpack for your boarding pass.

But as suggested, it is up to 99% correct on White Males (and up to 35% errors (or 65% accurate) for non-white / darker faces.

So, better, but not perfect.  

There have been attempts to scan large crowds at the SuperBowl and other events to try to identify terrorists.  It seems like it is getting better, but still not quite accurate.

So, let’s see if the technology gets better, and maybe, just maybe one-day facial recognition, with or without a haircut, with or without rapid weight gain or loss, – the technology will work for all people.  (Not sure if that is an invasion of our privacy or not!!!)

What do you think?

Bruce

Posted by Bruce White

Leave a Reply