A recent development from a respected researcher in the field of computer vision aims to address growing concerns about privacy associated with the increasing prevalence of video surveillance cameras.
The application of computer vision technology has proven to be highly beneficial, particularly to individuals or organizations tasked with the laborious process of analyzing extensive hours of surveillance video footage. For example, law enforcement agencies may utilize the technology to simplify labor-intensive searches with queries such as "Identify anyone wearing a red scarf over the last 48 hours".
With the ubiquity of video surveillance, Yogesh Rawat, Assistant Professor and researcher at the University of Central Florida's Centre for Research in Computer Vision (CRCV), is working on developing advanced software that integrates with the video cameras to address potential privacy concerns. The project is bolstered by $200,000 funding from the U.S. National Science Foundation's Accelerating Research Translation (NSF ART) program.
The primary purpose of Rawat's solution is the preservation of privacy in surveillance practices. He has developed video monitoring software that affords the recorded subjects their privacy by obscuring identifiable attributes such as faces and clothing, both during the recording process and in the surviving footage. This technology adds distortions to the RGB pixel component of the video feed—red, green, blue light colors—making them unrecognizable to human eyes.
Furthermore, this innovative software is being optimized for compatibility with edge devices; devices capable of operating independently of external servers, like drones and public surveillance cameras. The team at the CRCV is thus also tasked with developing technology capable of swiftly analyzing video feeds as they're received, necessitating the creation of algorithms able to process data at unprecedented speeds for compatibility with graphics processing units (GPUs) and central processing units (CPUs).
Rawat aspires to successfully implement his software across platforms, preserving privacy while also enhancing surveillance capabilities. He asserts that, "Automation lets us analyze vast amounts of footage, an impossibility for humans alone. Surveillance, while vital for society, often faces scrutiny over privacy intrusions. This initiative aims to marry the two—surveillance and privacy preservation."
In conclusion, computer vision, a remarkable domain in AI research, is amplifying its capacity to address societal needs and concerns. Rawat's pioneering work, a testament to this progression, balances the need for public surveillance with the growing demands for individual privacy in an increasingly interconnected digital landscape.
Disclaimer: The above article was written with the assistance of AI. The original sources can be found on ScienceDaily.
