By using facial recognition technology, a person’s face can be recognized. A facial recognition system identifies facial features based on a photograph or video. The information is compared to a database of known faces using a face matching algorithm. While identity verification using facial recognition can be useful, there are also some privacy concerns.
It is expected that the facial recognition market will grow by 15.6% from its current value of $3.8 billion market value. Numerous commercial applications drive the demand for facial recognition. From surveillance to marketing, facial recognition can be used for almost anything.
However, that’s when things get complicated. Those who place importance on privacy probably want to have control over how their data is used. The fact is that your “faceprint” is data.
The Basics of Face Recognition
Perhaps you are naturally adept at identifying faces. Perhaps you can easily identify your family and friends. You know the shape of their eyes, nose, mouth, and the color of their skin. Similar to facial recognition, an algorithmic version is also available. The computer sees data about the face rather than a face. It is possible to store and retrieve the data. One study by Georgetown University found that half of American adults have their faces stored in at least one facial recognition database that the police have access to.
We’ll now examine the process of facial recognition. The process begins by taking a picture or video of you. If you are with other people, the process will resemble that. If you appear in profile or directly ahead, the process will resemble that. If you appear in profile or directly ahead, the computer will recognize your face by its geometry. You also need to take into account the distance from your eyes to your chin and from your forehead to your chin. According to the software, 68 facial landmarks define your face. The result is that you can recognize your signature. This is done by comparing your facial signature to a database of known faces. It is estimated that at least 117 million Americans are represented in police databases. In May 2018, the FBI was reported to have access to 412 million facial images in search. Following a review of the images, decisions are made. Using facial recognition software, images from a database may match your faceprint.
Is Facial Recognition Accurate?
Those who criticize facial recognition argue that it could lead to false identification. Someone who breaks a store window during riots might get mistaken for someone who wasn’t there by the police? If this happened, how likely is it? There is no definitive answer. In an April 2020 test conducted by the National Institute of Standard and Technology, the facial recognition algorithm with the lowest error rate was 0.08. By comparison, the best algorithm had an error rate of 4.1% in 2014.
An article published in 2020 by CSI suggested that clear, static images of people, such as passport photos, can more accurately identify them. On the NIST’s Facial Recognition Vendor Test, these algorithms can achieve 99.97% accuracy when used this way.
Generally, accuracy rates in real life are lower, but this isn’t always the case. According to a CSI report, the error rate goes from 0.1% to 9.3% when high-quality mug shots and public photos are matched. Generally, this occurred if subjects were not looking directly at the camera or when shadows or objects were blocking their view.
An additional challenge comes with aging. Facial Recognition Vendor Test results indicated that middle-tier algorithms experienced significant error rates when matching images taken 18 years ago.
The Use of Facial Recognition
Policing and law enforcement rely heavily on facial recognition technology and the database built around it. It has been reported that police work contains a lot of comparisons between mugshots and facial databases to identify certain suspects.
Law enforcement agencies can use nearly any source to comb through the mugshot database, such as traffic cameras, closed-circuit televisions, social media, and photos taken by the police themselves. A police officer can take a photo of a driver or pedestrian using a smartphone, tablet, or another mobile device and then compare the photo to one or more facial recognition databases.
Significant events such as the Olympics, concerts, and other sporting occasions have included facial recognition to identify fugitives wanted for arrest. For the federal government, there are multiple facial recognition programs available. Among all of them, the FBI’s Next Generation Identification system is the one it uses most frequently. This database contains more than 30 million facial records.