Image Recognition can be dangerous

Facial Image Recognition software is real and here to stay.

Image recognition can jeopardise your privacy

Anyone with basic scripting skills can use your social media information to include you in facial image recognition software.
Quite scary stuff, showcases our Machine Learning Master Kevin Smeyers.

Video demo: watch your privacy being destroyed with image recognition

Is it time to change your profile picture?

Anyone with some scripting experience can build a simple Facebook Scraper to save all your public data. Such a scraper allows you to save your personal data like a profile ID, profile picture, profile name etc. in a database. Said database can be used in a facial recognition software application that can then be used for any coders’ desires.

How to create a facial recognition software application?

All you need is the incredibly hard skill of using Google to find your way to creating a working facial recognition application.

Public Facial Recognition API

A public git for the Facial Recognition API can be found here: https://github.com/ageitgey/face_recognition

The result of running the Facial Recognition API

It seems like the only people safe from being recognised are identical twins, as shown by Kevin Smeyers from ToThePoint.

Deep Learning HAAR cascade principle

The Deep Learning HAAR cascade principle used in facial recognition software is explained here: http://www.willberger.org/cascade-haar-explained/

The Deep Learning HAAR cascade principle.

Why create a facial image recognition software application?

A small-town grocery can show targeted promotions

Take the following example: You enter a small-town grocery and your face is registered by the security camera at the entrance. No big deal, right?

In the back room there is a computer running a script that recognises every person entering the place by name. Knowing what you like best thanks to your facebook likes, you’ll be shown a targeted promotion on a big screen at the entrance. All in a matter of microseconds.

The streets can be monitored

Not really the same as detecting people, but a usage of image recognition nonetheless: using image recognition to detect and count the trash to be collected in the streets. As shared by Maarten Sukel, AI specialist @ The City of Amsterdam.

Criminals on the loose can be tracked immediately

A more invasive example would be that of a criminal on the loose. The fugitive is being red-flagged by the authorities and when spotted by a public camera, he can immediately be localized by special police forces.

The public can be monitored

More creepy examples would include those used in China with the “social credit” point system where behaviour of people in the public space is being monitored.

The most creepy example we could find was the one of the Chinese “social credit” Point System.

Why you should protect your personal data

As shown in this blog, anyone can use the public information on your social media pages to include you in an image recognition software application. The only way to protect yourself from this, is ramping up your social media privacy settings.


Leave a Reply