"New Facial Recognition Technology Scans Your Ear"

The need for new authentication methods that do not require a person's full face to be visible has emerged in the post-COVID world of face coverings and amplified hygiene awareness. According to new research from the University of Georgia, people may soon be able to access their devices using their ears rather than their face or fingerprint. Thirimachos Bourlai, the lead author of the study, says the ear is one of the few body parts that remains relatively the same over time, making it a useful alternative to technology requiring face or fingerprint recognition. The ear recognition system developed by Bourlai's team correctly authenticates individuals with up to 99 percent accuracy, depending on the dataset and model used for testing. Ears, like fingerprints, are distinctive to each individual, as even identical twins' ears differ. An added benefit is that, with the exception of the earlobe, which drops lower over time, ears do not age in the same way that the face does. The ear recognition software functions similarly to face recognition software. When a person purchases a new phone, they must first register their fingerprint or face for the phone to recognize them. In order to get a complete "picture" of their fingerprint, new devices typically require users to place their fingers repeatedly over the sensor. Face-recognition technology relies on users moving their faces in specific ways in front of their camera for the device to effectively capture their facial features. The algorithm takes multiple samples of a person's identity, such as facial images or fingerprints, and logs them into the device while configuring a biometric device. When a user unlocks their device with a biometric, a live sample is taken and compared to the device logs, such as a picture of their face or, in this case, a picture of their ear. Bourlai's software evaluates ear scans and determines whether they are suitable for automated matching using an ear recognition algorithm. To test the software, he used various ear datasets with a wide range of ear poses. Bourlai tested his algorithm on two different datasets of ear images. In one dataset, system accuracy increased from 58.72 percent to 97.25 percent when compared to prior ear recognition software, while in the other, accuracy increased from 45.8 percent to 75.11 percent when compared to the baseline approach. This article continues to discuss the new ear identification technology developed by researchers at the University of Georgia. 

UGA Today reports "New Facial Recognition Technology Scans Your Ear"