In my last post, we looked at a few trends that face recognition has brought to us. Face recognition technology has increased the ability of law enforcement to catch criminals and for government agencies to track terrorists. In fact, many believe that the tragic events of September 11, 2001, could have been prevented in face recognition software had been in use at United States airports during that time.
An image shared by Digital Trends shows the terrorists who flew the planes into the World Trade Center as they entered the airport to board their flights. The article suggests that these men may have been recognized as potential terrorists by AI and face recognition. Along with enhancing the ability of law enforcement to track criminals and terrorists, smart shopping and identification verification are also trends brought on by face recognition. Here are a few more.
Face Recognition from Video
In its early stages, face recognition from images was the only type available. If someone had a still photo of you and could take another still photo of you, then the face recognition algorithm could recognize that both photos were of the same person. Moving pictures or face recognition video was much more complicated and therefore took a bit longer to achieve. This trend is important because it makes the ability for law enforcement to use this technology even greater.
Where it was once necessary for police to try to locate still photos of a crime scene and attempt to pick out a suspect from thousands of photos, now face recognition can locate a person from hundreds of hours of video. For example, at one time, police had to watch hours and hours of videos from convenience store robberies, and then scour over books and books of criminal photos just hoping to find a match and apprehend a suspect. Now, face recognition can match booking photos, Facebook photos, and other photos to an image on a video.
Face Recognition in Social Media
In 2014, Facebook announced its DeepFace program, which was basically face recognition on Facebook. With an accuracy of over 97%, Facebook’s face recognition system can determine whether or not 2 photos are of the same person, according to Gemalto. This is almost as accurate as human recognition. People now use this face recognition technology to tag one another in photos every day. Of course, for now, Facebook gives a user the ability to opt-out of using facial recognition technology or allowing his or her face to be recognized or tagged.
To those who kept up with face recognition trend and abilities, Facebook’s ability to identify photos with over 97% accuracy was an amazing feat. In 2015, however, Google raised the bar. Face recognition on Google was known as “FaceNet,” and tested with an accuracy of 100% in a reference test called “Labeled Faces in the Wild.” That technology has been incorporated into Google Photos, making it possible for face recognition technology to sort through your photos and automatically tag people in them for you. An unofficial online version of this technology has been released online as well. It is called OpenFace.
Face Recognition Isn’t Foolproof…Yet
Those are a few of the face recognition trends that are already in place. Many more are expected as products such as Amazon’s Rekognition becomes available. Apparently, Amazon still has a few kinks to work out, however, since Newsweek reported in 2018 that its face recognition system had incorrectly identified 28 members of Congress as criminals by matching their photos with mugshots.