This summer the FBI NGI, Next Generation Identification, becomes reality, according to The Verge today. However, questions surrounding the accuracy of the FBI facial recognition data base have surfaced.
Facebook may be better in its facial recognition database at 97 percent accuracy compared to the FBI at 85 percent.
This past April the Electronic Frontier Foundation had posted a piece about the upcoming 50 million plus facial recognition views to be placed into the FBI NGI base. Despite the extensive fingerprint base and individual states gathering data into the computer, the Electronic Frontier Foundation found gaps in the FBI program. The gaps are a cause of concern.
Compare that to Facebook's DeepFace system, presented at the IEEE Computer Vision last month, Facebook has an advantage in providing 97 percent accuracy due to its data base collection of members. The FBI has a much harder uphill battle due to the types of people that it wants on its facial recognition base.
According to Shahar Belkin CTO of FST Biometrics there is a difference between a human brain and a computer brain that makes it difficult to compare to high accuracy. FST has a system of facial recognition that works with willing participants and subjects who it can capture. In addition it is interesting that FST has an additional layer of verification of a profile recognition that includes height and gait-tracking.
The FBI system does not have extra layers of verification so Belkin believes that, "What will kill these systems is the false-accept rate,” and adds, "I don't believe we'll see a solution for that in the next five to ten years."
Another problem is that mounted cameras in public places take pictures at an angle which will lead to false reading or impossible to identify with any accuracy. The varying degrees of difficulty will make the program become untenable. Most of the people that the FBI want to capture are avoiding facial recognition.
Facebook has already 250 billion photos by comparison to the FBI 50 million. The Facebook photos are mainly willing participants and posing for the Facebook placement. It is easy for Facebook to identify and match facial recognition because its data base has connected people with their friends and photos.
The issue may be resolved in court since Facebook is in legal battle with the Manhattan district attorney over how broadly Facebook data can be collected by the prosecutors for a court case. If Facebook loses then law enforcement will have its build of faces and reach into Facebook. That end result is what has been feared about the countless collection of personal data and revealing photos on Facebook.
Facebook has other privacy issues with the FTC over misuse of personal data related to its 1.3 billion users. The Electronic Privacy Information Center filed a complaint to the FTC. In its complaint the privacy group called, "Facebook’s secretive and non-consensual use of personal information to conduct an ongoing psychological experiment on 700,000 Facebook users” a violation in psychological intrusion.
The FTC and the Manhattan court response may give law enforcement and the FBI the edge that its needs in the program of NGI. It is only a matter of time that the FBI will develop more sophisticated methods of facial recognition.
The NSA far more clouded in secrecy has been collecting images from the Internet according to myfox.com online today. This makes Big Brother chilling in real time a rapid descent into a situation of no privacy.