Faculty Mentor Information
Dr. Jyh-Haw Yeh (Mentor), Boise State University
Additional Funding Sources
Supported by National Science Foundation Award #2244596 REU Site: Cloud Computing Security and Privacy.
Presentation Date
7-2024
Abstract
As younger generations gain easier access to the internet, the need for effective online age verification becomes increasingly critical to ensure age-restricted activities remain accessible only to those who meet legal requirements. Modern online age verification often suffers from either intrusive personal data collection or ineffective blockage. Accurate age verification requires government ID or sensitive information to be exposed, while methods that don’t collect any information from users aren’t suitable for preventing access. This research aims to find an online age verification method that is both accurate and respectful of user privacy using AI and machine learning. The most commonly used metric for AI-based age estimation is facial images, which have proven relatively accurate for determining exact age. This method holds promise for balancing privacy with age gates as it does not require a user's name, address, or other sensitive information—something especially important for minors. To further protect user privacy, we explore age estimation using low-quality facial images, which provide less detail for analysis by both people and computers. If results with low-quality images are shown to retain accuracy, they could pave the way for future age estimation software.
Privacy-Preserving AI-Based Age Verification Using Low Quality Facial Images
As younger generations gain easier access to the internet, the need for effective online age verification becomes increasingly critical to ensure age-restricted activities remain accessible only to those who meet legal requirements. Modern online age verification often suffers from either intrusive personal data collection or ineffective blockage. Accurate age verification requires government ID or sensitive information to be exposed, while methods that don’t collect any information from users aren’t suitable for preventing access. This research aims to find an online age verification method that is both accurate and respectful of user privacy using AI and machine learning. The most commonly used metric for AI-based age estimation is facial images, which have proven relatively accurate for determining exact age. This method holds promise for balancing privacy with age gates as it does not require a user's name, address, or other sensitive information—something especially important for minors. To further protect user privacy, we explore age estimation using low-quality facial images, which provide less detail for analysis by both people and computers. If results with low-quality images are shown to retain accuracy, they could pave the way for future age estimation software.