These days, an AI system has learned to reliably forecast both age and ethnicity, making facial recognition technology contentious.
Age and gender are two of the most important facial characteristics. Since they are so fundamental to social interactions, estimating them from a single-face image is critical in intelligent applications like access control, HCI, law enforcement, marketing intelligence, and visual surveillance, among others.
Researchers have created an algorithm that accurately analyzes face characteristics, including wrinkles, furrows, and spots. The Institute for Neural Computation at Ruhr-Universität Bochum (RUB) has developed a system that uses AI to determine a person’s age and ethnicity precisely.
AI Algorithm Identifies Age & Ethnicity
The age and ethnicity of faces can now be determined by an AI system, albeit experts are unsure of the method’s exact mechanism. Due to its extensive adoption by law enforcement and the potential for racial prejudice, AI face recognition technology has recently come under fire.
Amazon announced a police limitation on the usage of its face recognition technology in the wake of the Black Lives Matter demonstrations until better guidelines are implemented. However, concerns about using AI-based face recognition existed even before the most recent innovations. A Chinese business created a facial recognition system that could recognize a face while hiding behind a mask during the COVID-19 epidemic. Meanwhile, companies like Clearview AI have been developing face recognition algorithms that may risk internet users’ privacy. Despite these instances, it is anticipated that the technology will be widely employed in the future, particularly in crowded urban areas.
According to Tech Xplore, engineers from Ruhr-Universität Bochum (RUB) in Germany developed the most recent AI innovation. The system reads facial characteristics of aging, such as wrinkles, furrows, and spots. The system is also capable of determining a face’s ethnicity in addition to its age. Given that many of the existing AI facial recognition algorithms are less capable of reliably judging non-white faces, this characteristic is all the more important.
How AI Learns To Identify Age & Ethnicity
The method in question is an eleven-level hierarchical deep neural network. Researchers supplied the AI with hundreds of photographs with a range of ages and ethnicities as well as information on the ages and ethnicities of each individual in the pictures to train it. Researchers gave the machine the freedom to decide which ethnic and facial characteristics to examine when predicting recognition. This included facial wrinkles that deepen gradually with age to aid in age identification. The algorithm maintained those particular characteristics and disregarded others that frequently fluctuate between faces, such as eye color or nose shape, in accordance with the design.
The researchers have not yet been able to pinpoint the traits the system uses to distinguish between ages since the algorithm taught itself to read faces. Even still, the outcome is astounding, especially given that the algorithm typically underestimates age by no more than three and a half years. An accomplishment was allegedly superior to the human brain, our current best system. The likelihood of correctness was found to be better than 99 percent for ethnicity. While the accuracy in identifying age and ethnicity is impressive, there is a transparency issue because engineers aren’t aware of how the AI system arrives at its findings.
