Keeping Your Hotel Room Secure with Facial Recognition
Photo Source: Pixabay
Most of us are old enough to remember at a hotel using a metal key from the front desk to open our rooms. Today, hotels use a card with a magnetic strip to open the room lock. These keys help ensure the safety of our rooms and belongings while you’re away from the hotel room. For improved security, hotels moved to the magnetic key card because, unlike metal keys, the locks can be changed anytime with the press of a button, making the rooms more secure, as opposed to a metal key, where it takes a lot longer to replace a lock in the event a key has been copied and the lock has been compromised. However, like a metal key, the magnetic keys can be stolen or copied, still rendering hotel security imperfect. With the help of artificial intelligence, the next level of hotel security has arrived and involves one of the most unique keys in the world—your face.
Facial recognition computers can now use a person's face as the key to check them into the hotel without having to wait in line to check in at the front desk, and the guest's face serves as the key to enter their hotel room. According to Hotel Technology News Vinpearl Nha Trang, a five-star hotel brand in Vietnam, has implemented facial recognition to allow guests to check in in only three seconds, automatically open doors, and automate payment for services. (hoteltechnologynews.com)
Artificial intelligence has dramatically accelerated the development of computer facial recognition. In the Handbook of Facial Recognition, Stan Z. Li and Anil K. Jain describe various approaches computer scientists have used over the years to develop computer facial recognition. Earlier methods used explicit programming to define faces using specific points such as nose, eyes, and chin to make a measurable map of the face. Features such as the distance between the eyes and the nose's length can then be used to search a database of photographs with known faces for a match. This technique works well in very controlled situations and when the faces and pictures line up nicely. Still, in real life, people tilt their heads, lighting changes all the time, and often only partial faces can be seen, rendering explicit mapping ineffective at accurately recognizing faces. However, with artificial intelligence, computer facial recognition has dramatically improved in recent years using something called deep learning.
With deep learning, the computer programmer does not tell the computer what a face is; instead, she programs the computer to learn from examples much the way that we teach children. For example, when teaching a child to recognize a horse, we show them many pictures of horses and repeat over and over again, "This is a horse. This is a horse." We don't tell the child to count the legs, look for hooves, big eyes, a long tail, and a long mane to identify a horse. In a way, deep learning learns like a child, and it has transformed the quality and accuracy of facial recognition to human-level ability. A Chinese company, Baidu, published a paper that claims that their artificial intelligence-enabled facial recognition system called Facial Recognition via Deep Embedding has a 99.7% accuracy in facial recognition, which is better than a human.
Deep learning made facial recognition accurate and, just as importantly, affordable. Beyond hotels, many institutions have embraced facial recognition technology to manage security and enhance their services. According to an article at Research Briefs.com, "Facial Recognition Is Already Here: These Are The 30+ US Companies Testing the Technology." (cbinsights.com) Walmart uses facial recognition to track shoppers to prevent shoplifting. Automotive manufacturers such as Ford and Chrysler are testing the technology for car entry and driver identification, and Cigna uses facial recognition in place of a signature at its health insurance subsidiary in China. Casinos, law enforcement, border control, and even iPhone now offers facial recognition in place of a password to improve security and convenience.
Facial recognition represents a considerable step forward in security. Still, creative thieves have already figured out ways to trick computer facial recognition into giving them entry and breaking into people's rooms, compromising their safety and property. Facial recognition computers focus on verifying a face's identity, but the computer does not know anything about whether the face it sees is real. Because of this, thieves figured out one of the simplest methods of fooling facial recognition, using a picture of their victim, to be extremely effective. They can take a picture of their victim, print it out, and present the image to the facial recognition camera to fool the system, gaining entry to places that should have only been accessible to the victim. More sophisticated attacks involve making lifelike masks or presenting a video of the victim to the facial recognition system.
In the continuous game of cat and mouse between security and criminals, computer scientists have developed new techniques for the facial recognition system to figure out if the face it sees is alive and not a picture, mask, or video. This technology goes by the name, liveness detection, or the ability to discern whether a face is actually alive and not a photo or mask. In an article published in the International Journal on Information Theory, titled "An Overview of Face Liveness Detection," the authors describe several techniques to detect a face's liveness. Unlike a real face, a static picture has irregular skin texture, which the facial recognition system detects using mathematical comparisons of different parts of the face. More sophisticated systems use 3D cameras to determine if the scanned face is authentic or just the flat image of a picture or a video. 3D cameras can see the difference between a nose in a photo and the nose on a real face. Combating lifelike masks requires more active security, such as requests from the system to make real-time facial gestures to prove liveness.
Physical keys and now magnetic key cards currently protect hotel guests in their rooms. Still, increasingly facial recognition technology will use one of the most unique keys of all, the face. However, hackers and thieves have found many ways to fool the system with pictures, videos, and masks. In response to these workarounds, new techniques to prove liveness now accompany facial recognition to help the system determine the real faces from the fake. Make sure if your hotel uses facial recognition to safeguard your room that it has the latest liveness detection, or you may not be as safe as you think. However, it may be a little while before it eclipses the key card.
Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.
You can buy his book on Amazon in paperback and in kindle format here.