ProLinks are back after a long silence thanks to new ideaton sessions we have started in our consulting team at Innovatrics. Here is what happened.
Consultant works more or less like a one man army – visiting customers and helping them solve whatever issue they have related to our products and projects. We therefore often get out of sync on what is happening with other projects, clients, competitors, tools or generic news.
So, I have promised that I will try to keep my team up to date with the latest news. And while I’m at it, lets keep up to date everyone who is interested.
Because at Innovatrics we focus on biometrics, for now most news will be focused on that. Of course, now and than we discover something that is really useful for anyone, so there will be also a section for stuff like that.
So from now on, every two weeks I will try to compile what I have found on the Internet about biometrics. That’s enough for an explanations, lets jump in on the news.
ID Talk Podcast
We live in an era of podcasting and I don’t know about you, but I love to consume information this way. That’s why I’m glad that FindBiometrics and Mobile ID World have started publishing interesting discussions with people from biometric industry.
Feature that allows you to group your photos by faces on them, which was available in some countries for a long time, is spreading. The claim, that everything is stored locally and fully in your ownership is really interesting.
Blacklisting Chinese facial recognition vendors
The US is cracking down on Chinese companies which provide facial recognition algorithms. This can shuffle the market, both due to sanctions and also due to possible removal of sanctioned algorithms from NIST FRVT benchmarks.
Facial recognition vendors in trouble
Most facial recognition vendors are in trouble anyway, as one of the most used free databases MegaFace contains images protected under strict BIPA regulation, which allows some US citizens to sue them, if they have used this dataset for algorithm development without their permission.
Datasets to grab
Mentioned in the article above is project Megapixels, which links several interesting datasets for grab. Be careful about the regulations though :).
Facial algorithms vulnerable to infrared lights
Researches have demonstrated that they can fool facial recognition algorithms using simple infrared lights. Interesting fact is, that this works both for avoiding face detection and also for impersonating somebody else.
Facial algorithms vulnerable to morphing
As if infrared is not enough, it seems that facial recognition algorithms have difficulties detecting morphed faces and deep-fakes.
Helping visually impaired
While cameras used by police officers which allow them to identify everybody in real-time can be scary, the same principle can help visually impaired to recognize their friends and family.
California is killing passwords
New regulations in California called CCPA can help boost biometric industry. This is happening despite the war on facial recognition waged by the same state.
Three reasons to worry
NYT has published short op-ed about the possible issues with facial recognition, including racial biases and sending wrong people to jail. Valid points, though sometimes overblown.