Facial recognition – my phone is better than I am

A facial composite produced by FACES software

A facial composite produced by FACES software (Photo credit: Wikipedia)

I’ve never been particularly good at recognising faces or at putting names to faces. Unfortunately, this is compounded by the fact that I have to meet a lot of people as part of my job. I used to be really envious of my optician who remembers every single customer by name. Either he doesn’t have many customers or more likely he is very good at recognising people.

Facial recognition seems to be everywhere now. Most smartphones have the technology built-in. My wife’s camera has it. Even Facebook has it, and it tends to be pretty accurate too. Facial recognition has come a very long way since I first played with it 15 years ago.

We were doing some research for some US department of something or other on whether facial recognition and crowd recognition software could be used to tackle such diverse issues as crowd control, anti terrorism and to pick out pickpockets in a large group of people.

The blurb behind the recognition engines was impressive. Apparently, the software could pick out the telltale movement signatures of someone who was up to no good or by looking at a crowd, predictions could be made as to the underlying mood of the crowd and how likely they were to become hostile.

We had to take their word for some of these features, because our lab lacked both a large crowd and a friendly felon to act shiftily. We all took it turns to pretend to act shiftily, but either the machine was too clever to be fooled or we just didn’t have it within us act nefariously enough.

We could, however, test out the facial recognition engine. We each took it in turns to have our photograph taken to give the software a library of faces to choose from. All in all, we had about half a dozen photos. Once we had the photos, we took turns to pose in front of the camera to see how the facial recognition engine worked in practise.

All in all, it wasn’t too bad at recognising a full face image, getting it right probably 3 out of 4 times. The trouble was the time the processing took. If you were pointing your camera at a crowd of people trying to pick out a known felon, your crowd would be half a mile down the road before the software had made it through half the crowd. The processing power simply wasn’t enough back then.

I assume that with the increase in processing power in the last 15 years, such software is viable these days and in common usage in airports and at large stadium events around the world. Even though it all sounds a bit big brother, I think it’s a good thing if it improves the safety of the general populace.

I just wish they would hurry up and install facial recognition into my spectacles so that I might be able to recognise someone other than my optician.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s