If you’re lucky enough to have a ticket to an event at Madison Square Garden in New York (say, an NBA Finals game), one aspect of your visit will be having your face scanned by a facial recognition system.
Large event spaces are increasingly using this technology. Some, like the Garden, use it for surveillance purposes and others, like Citizens Bank Park in Philadelphia and Oracle Park in San Francisco, to offer visitors optional ticketless entry.
Adoption of facial recognition technology is increasing and becoming more prevalent in daily life, from public buses to public buildings. The Transportation Security Administration has implemented the latest facial recognition technology at security checkpoints at numerous airports. The agency says the new system will be used in cities in the United States that will host 2026 FIFA World Cup soccer matches.
The growing use of facial recognition has amplified concerns about accuracy and bias. But in my research on facial recognition technology at the University of Dayton Vision Laboratory, I found that advanced deep learning models have made facial recognition systems more accurate and reliable. AI models, trained on hundreds of millions of face images, are over 99% accurate in controlled environments such as mobile phones, airports and border checkpoints.
Basics of facial recognition
Facial recognition involves three steps: locating a face in an image or video frame, creating a facial fingerprint that catalogs salient features (including face shape and landmarks such as eyes, nose, and mouth), and recording skin texture. It then compares the facial print with those in a database, which may be inside a smartphone or in a bank or hospital, to verify a person’s identity or allow access.
In the physical world, these systems are faster and easier than requiring people to show identification documents. In the online world, they are easier than entering a username and password. Facial recognition also significantly reduces the possibility of forgery or fraud compared to ID cards or passwords.
Improvements in technology come from a variety of research projects. FaceNet, a deep learning model developed by Google, has improved the recognition of faces that are partially covered or hidden in images. DeepFace, a landmark AI-powered facial recognition system developed by Facebook AI Research, achieves the same high level of verification that humans display.
NeoFace, a high-precision AI-powered algorithm developed by NEC, is integrated into Mobile Fortify, the mobile facial recognition system used by U.S. Immigration and Customs Enforcement to identify people.
Reduce false positives and negatives
Real-world conditions such as poor lighting, difficult viewing angles, extreme facial expressions, obscuration by face masks or sunglasses, and poor image quality can still hinder performance and lead to faulty identification. False positives and false negatives are the two main errors. False positives occur when a person is incorrectly matched with a different person in a database. False negatives occur when an individual is not found in a database, even though their image does exist there.
False positives are more critical in security and safety applications. They can lead to unfair accusations, discrimination or arrest. In 2025, a 50-year-old woman in Tennessee was arrested and jailed for six months due to an AI-powered facial recognition system that incorrectly linked her to a bank fraud investigation in North Dakota. False negatives can lead authorities to deny services to people who qualify to receive them.
Accuracy can suffer if models are trained on data that does not reflect real-world demographics. A 2025 study showed that systems trained on public databases missing people with darker skin tones lead to lower recognition accuracy. This type of unintentional bias in training data can lead to misidentification of women, people of color, and the young and old. A report found that facial recognition systems used by 42 US government agencies falsely identified African American and Asian faces 10 to 100 times more often than white faces, in some cases leading to wrongful arrests.
Accuracy also deteriorates when people wear a lot of makeup and for young children and older people because their characteristic features tend to change more rapidly than adults of other ages. Balancing data sets by collecting more representative images based on age, gender, and ethnicity, and updating databases frequently can improve accuracy and produce fairer results.
Adjusting images before sending them for comparison (for example, changing brightness levels) can also improve accuracy. People squint when they are in the dark or in very bright light. Advanced processing software can mimic this human feature to improve the facial recognition system’s ability to extract facial features from the image.
A complete face from partial data.
Humans are good at identifying a person even if part of their face is covered by sunglasses or a mask. The brain assigns more importance to the exposed details. If facial recognition programs can learn to do the same, that would reduce false positives and false negatives, even when cameras only capture part of a face.
Facial dynamics can also help. It may be difficult for someone to recognize a high school friend they haven’t seen in years, but if the old friend smiles, that change of expression can immediately enhance the memory.
Researchers are developing a facial recognition method to do this, known as volumetric directional patterning. It captures subtle movements of facial muscles, as well as the blinking of eyelids, in consecutive frames of a video. It tracks how facial landmarks change over time, as well as the context in which a face is viewed, which can improve recognition accuracy.
Researchers are also creating more precise AI-powered three-dimensional systems that can capture the precise geometry of a face, including features such as the contours of the eye socket, nose, and chin. This type of work could lead to anti-spoofing techniques that prevent facial recognition systems from falling for fake faces generated by computers and their human operators.
Fewer mistaken identities
Privacy and cybersecurity issues and persistent issues of bias aside, one thing is clear: facial recognition technology is improving. And that promises fewer mistakes and fewer serious consequences that accompany them.
Vijayan Asari is a professor of electrical and computer engineering at the University of Dayton..
This article is republished from The Conversation under a Creative Commons license. Read the original article.