IEEE SPS SCV - Leveraging Image Processing for Forensics and Privacy in the Digital Age
This talk explores the intersection of image processing and computer vision with crucial forensics and privacy challenges, shedding light on innovative methods and inherent vulnerabilities. We begin by examining the potential of consumer-grade cameras for capturing detailed 3D surface structures at a microscopic level, enabling extracting the "fingerprint" information for the high-accuracy, low-cost authentication of important documents and product packaging. Building on the theme of information extraction, we delve into how indoor camera images can inadvertently reveal location information, offering a powerful tool for the potential capture of child exploiters. Finally, we transition to analyze the vulnerabilities within federated learning systems. We demonstrate how training images can be silently extracted by a curious but honest server or from participating clients, raising significant concerns about data privacy for the potential deployment of federated learning systems.
Date and Time
Location
Hosts
Registration
- Date: 18 Oct 2024
- Time: 12:00 AM UTC to 01:30 AM UTC
-
Add Event to Calendar
- Starts 06 October 2024 07:00 AM UTC
- Ends 18 October 2024 01:00 AM UTC
- No Admission Charge
Speakers
Chau-Wei Wong of North Carolina State University, Raleigh, NC, USA
Leveraging Image Processing for Forensics and Privacy in the Digital Age
This talk explores the intersection of image processing and computer vision with crucial forensics and privacy challenges, shedding light on innovative methods and inherent vulnerabilities. We begin by examining the potential of consumer-grade cameras for capturing detailed 3D surface structures at a microscopic level, enabling the extraction of the "fingerprint" information for the high-accuracy, low-cost authentication of important documents and product packaging. Building on the theme of information extraction, we delve into how indoor camera images can inadvertently reveal location information, offering a powerful tool for the potential capture of child exploiters. Finally, we transition to analyze the vulnerabilities within federated learning systems, demonstrating how training images can be silently extracted by a curious but honest server or from participating clients, raising significant concerns about data privacy for the potential deployment of federated learning systems.
Biography:
Chau-Wai Wong received his B.Eng. degree with first-class honors in 2008 and an M.Phil. degree in 2010, both in electronic and information engineering from The Hong Kong Polytechnic University. He completed his Ph.D. in electrical engineering at the University of Maryland, College Park, in 2017. He is an assistant professor at the ECE Department, the Forensic Sciences Cluster, and the Secure Computing Institute at NC State University, USA. He was a data scientist at Origin Wireless, Inc. His research interests include multimedia forensics, machine learning, statistical signal processing, and video coding, with a recent focus on federated learning and generative models. Dr. Wong is a recipient of a top-four student paper award and the NSF CAREER award. He is an elected member of multiple technical committees and has been a regular tutorial speaker at top international conferences. He was involved in organizing the third edition of the IEEE Signal Processing Cup in 2016 on electric network frequency forensics. [Webpage: https://ncsu-wong.org/].
Address:Raleigh, United States