Professor Dr Baek-Young Choi: The Eyes Have It – liveness detection of mobile authentication & deepfake detection

IEEE NJ Coast PACE SIGHT Group, together with Computer Chapter and cosponsors present

Professor Dr Baek-Young Choi.

Title: The Eyes Have It

Abstract: In this talk, we will discuss how eyes can be used for liveness detection of mobile authentication and deepfake detection.

As the need for contactless biometric authentication becomes more significant during COVID-19 and beyond, the popular biometric authentication method for mobile devices, iris detection, and facial recognition confronts various usability, security, and privacy concerns, including mask-wearing and various Presentation Attacks (PA). Specifically, liveness detection against spoofed artifacts is one of the most challenging tasks as many existing methods cannot conclusively assess the user’s physical presence in unsupervised environments. Even though several methods have been proposed for tackling PA with motion challenges and 3D mapping, most require expensive depth sensors and fail to detect sophisticated 3D reconstruction attacks. We present a software-based face PA Detection (PAD) method named, which creates challenges and detects meaningful corneal specular reflection responses from human eyes. To detect human liveness, our system generates multiple screen image patterns as a challenge, then captures the response of corneal specular reflections using a frontal camera and analyzes the images using lightweight Machine Learning (ML) techniques. Liveness detection system components include challenge pattern generation, reflection image augmentation (e.g., super-resolution), and ML-based analyses. We have implemented the system as Android, iOS, and web apps. Our extensive experimental results show that our scheme achieves liveness detection with high accuracy at around 200 ms against various types of sophisticated PAs. The liveness detection can be applied for multiple contactless biometric authentications accurately and efficiently without any costly extra sensors nor involving users’ active responses.

The second part is about deepfake detection through eyes and environment. Deepfake techniques presenting AI-generated fictitious facial images of people can negatively influence the authenticity of online information. Starting as benign mesmerizing memes, they can be used malignantly to originate deception, manipulation, persecution, and seduction, defying societal quality and human rights. However, due to the recent development of sophisticated deepfake generation technologies, it is getting harder to distinguish counterfeit images. Thus, instead of relying on a single aspect of the visual features, we detect various features from the specular reflection images such as color components, shapes, and textures to check the coordination with the surrounding environmental factors such as indoor/outdoor, bright/dark, backgrounds, and light strength. We have conducted extensive experiments to evaluate the performance of our method using various input parameters and advanced Deep Neural Network (DNN) architectures on multiple public DeepFake datasets. The empirical results show that our technique achieves high accuracy (99.0%) in detecting sophisticated deepfake images.

Short Bio:

Dr. Baek-Young Choi is a Professor at the University of Missouri – Kansas City (UMKC). She received her Ph.D. degree in Computer Science and Engineering from the University of Minnesota, Twin Cities. She published three books on network monitoring, storage systems, and cloud computing. She has been a faculty fellow of the National Aeronautics and Space Administration (NASA), U.S. Air Force Research Laboratory’s Visiting Faculty Research Program (AFRL-VFRP) and Korea Telecom’s – Advance Institute of Technology (KT-AIT). She is an associate editor for IEEE Consumer Electronics Magazine and was an associate editor for IEEE Internet-of-Things Journal, Springer Journal of Telecommunication Systems, Elsevier Journal Computer Networks. Her research interests generally lie in the broad area of networking and communications, with specific emphasis on Internet-of-Things, cybersecurity, and smart city technologies. She is a senior member of ACM and IEEE, and a chair of IEEE Women in Communications Engineering (WICE).

Baek-Young Choi, Ph.D. Professor of Computer Science School of Science and Engineering University of Missouri-Kansas City

Speaker(s): Professor Dr Baek-Young Choi,

Agenda:
7pm Meet and Greet the Visiting Professor Dr Baek-Young Choi.

Presentation.

Introduction to Women In Communications Engineering / WICE & WIE Connections.

Say hello to colleagues.

Summary and conclusions.

Virtual: https://events.vtools.ieee.org/m/347211