Privacy and Face Recognition Tech

Facial Recognition

Topic:

     Eyewitness identification of criminal suspects has been a common place investigation tool for police departments across the country. This method of identification is most often implemented through the use of police lineups or by having victims or witnesses review photos of several individuals. These methods of identification are slowly being replaced, however, with “virtual line-ups” in which, instead of a human identifying a suspect, an algorithm using facial recognition technology to do so instead.

Although the algorithms vary, the general idea behind facial recognition is a fairly straight forward. First, an algorithm must identify a person’s face within an image it has been presented, a process known as facial detection. The algorithm then works to rotate and align the face with an image from its database so that it can compare the faces in the same orientation. The facial recognition software then tries to determine if the face in the image and the face from the database represent the same individual. Facial recognition software does not produce a binary “yes” or “no” answer, however, but instead rates the probability that the faces do in fact  represent the same person.

Facial recognition software is trained to improve results. During training, an algorithm is given pairs of face images of the same person. Over time, the algorithm learns to pay more attention to the features that were the most reliable signals that the two images contained the same person. This method of training can be problematic, however, if the training set skews towards a certain race. This can cause the algorithm to be better at identifying members of that one racial group over others. In 2006, a NIST competition compared the accuracy of facial recognition algorithms developed in East Asia vs those developed in the US and Europe. From this competition it was found that the algorithms developed in East Asia performed better on East Asians while those developed in Europe and the U.S. performed better on Caucasians. This has led to concerns about potential racial biases in facial recognition systems.

Privacy Concerns Surrounding Facial Recognition

The increasing use of facial recognition technology by law enforcement has led to serious concerns about privacy infringement. The article suggests that the use of face recognition technology may curtail our First Amendment rights. While The Supreme Court has agreed that the mere surveillance of speech is insufficient grounds for a First Amendment violation, it has also held that the First Amendment protects the right to anonymous speech and association. Thus, the article argues, if facial recognition is applied in a situation, such as a protest, where people are voicing their opinions while not necessarily revealing their identities, it may violate their First Amendment rights.

Facial recognition is also a concern when considering the fourth amendment. Today, real-time face recognition is generally computationally expensive and not instantaneous. However, the technology has begun to see use among law enforcement agencies with the LAPD being the first US police department to openly acknowledge its use of real-time facial recognition technology. Here, we can begin to see how facial recognition creates opportunities for tracking which other biometric identifiers, such as fingerprints, do not. While warrants and other protections are often applied to traditional means of tracking, such as placement of a GPS device on a suspect’s car, courts and legislatures have not applied these protections to real-time facial recognition. While the Fourth Amendment gives us protections against “unreasonable searches and seizures,” it is unclear whether face recognition constitutes a “search.” In the past, courts have limited geolocation tracking under the fourth amendment but as of yet, facial recognition has not been limited such.

Potential Benefits of Biometrics

Today, law enforcement agencies mainly use face recognition for two purposes, face verification and face identification. Face verification seeks to confirm someone’s claimed identity while face identification seeks to identify an unknown face. One example from the article which demonstrates the successful use of face recognition is as follows: “Neil Stammer was a fugitive wanted for child abuse and kidnapping who had evaded capture for 14 years…Then, in 2014, a State Department official with the Diplomatic Security Service ran the FBI’s wanted posters through a database designed to detect passport fraud—and got a hit for Kevin Hodges, an American living in Nepal.” It was Neil Stammer, who was then arrested and returned to the US to face charges. In such a case, it is clear that face-recognition can have significant benefits, especially when used to aid law enforcement.

My Conclusion

            After reading the article it is clear that the use of face recognition by law enforcement has its associated benefits and drawbacks. I would recommend that Utah law enforcement make use of facial recognition technology, but with the following stipulations. First, I do not support real time facial recognition because I feel that its use has many associated privacy concerns, such as the possibility of tracking, which other uses of the technology do not. Also, I don’t believe the technology should be used on images or videos of the general public but instead only used on images featuring probable suspects in ongoing cases. The alternative, I feel, would be a suspicion-less search and illegal under the Fourth Amendment.

Relevant Links

Link to the Article: https://www.perpetuallineup.org/sites/default/files/2016-12/The%20Perpetual%20Line-Up%20-%20Center%20on%20Privacy%20and%20Technology%20at%20Georgetown%20Law%20-%20121616.pdf