Advancing Privacy in Surveillance: UCF’s Innovative Software Solution
In an era where video surveillance is omnipresent, the discussion around privacy has become more pertinent than ever. Addressing these concerns, Assistant Professor Yogesh Rawat from the University of Central Florida (UCF), affiliated with the Center for Research in Computer Vision (CRCV), is at the forefront of creating pioneering software designed to protect individual privacy in surveillance footage. This groundbreaking work is propelled by a substantial $200,000 grant from the U.S. National Science Foundation’s Accelerating Research Translation (NSF ART) program.
Computer vision technology has significantly transformed how surveillance footage is analyzed, offering the potential to quickly identify individuals through specific queries. Yet, this capability raises legitimate concerns over privacy. “Automation allows us to watch a lot of footage, which is not possible by humans,” mentions Rawat, emphasizing the dual need for surveillance and privacy preservation within society.
“This development will enable surveillance with privacy preservation,” states Rawat, as he describes the core ambition behind his project.
The innovative software Rawat is developing aims to obscure identifiable elements within videos, such as faces or unique clothing patterns, both in recorded footage and in real-time processing. This is achieved by adding subtle perturbations to the RGB pixels within the video feed, effectively making it difficult for the human eye to recognize specific details, while still maintaining the overall context of the scene for analysis.
“Our focus is on any visually interpretable identifiable information,” Rawat elaborates. This includes various personal characteristics, such as facial features, height, hair color, and body shape. The goal is to ensure these details remain private and unidentifiable in the footage.
One of the key challenges Rawat and his team face is optimizing the technology for edge devices, such as drones and public surveillance cameras, which operate independently of external servers. These devices require rapid processing capabilities to analyze video feeds in real-time without significant computational resources. “We want to do this very efficiently and very quickly in real time,” says Rawat, highlighting the importance of speed and computational efficiency in implementing this privacy-preserving software.
The NSF ART grant will further enable Rawat to explore potential applications of this technology across various sectors, including nursing homes, childcare centers, and public surveillance systems. His project is among the initial endeavors supported by a larger $6 million grant awarded to UCF, aiming to fund additional innovative projects over the coming years.
Rawat’s work is a continuation of the UCF CRCV’s legacy in computer vision research, building upon projects initiated by CRCV founder Mubarak Shah and researcher Chen Chen. Their previous efforts have laid the groundwork for analyzing untrimmed security footage, scaling artificial intelligence models for efficiency, and developing software capable of detecting multiple subjects and actions within video streams. This body of work has attracted over $7 million in funding from various prestigious organizations, including IARPA and the U.S. Combating Terrorism Technical Support Office.
Motivated by a desire to enhance societal well-being through technology, Rawat concludes, “I’m really interested in understanding how we can easily navigate in this world as humans. Visual perception is something I’m very interested in studying, including how we can bring it to machines and make things easy for us as humans and as a society.” This project not only signifies a leap forward in privacy-respecting surveillance technology but also showcases UCF’s commitment to harnessing computer vision for the greater good.
