A team of researchers from the Luddy School of Informatics, Computing, and Engineering has been named runner-up for the 2020 Privacy Protection Prize awarded by France’s Commission Nationale de L'informatique et des Libertés (CNIL) and National Institute for Research in Computer Science and Automation (Inria) for their paper on a method using computer vision techniques to protect bystanders’ privacy in photos.
The study, “Automatically Detecting Bystanders in Photos to Reduce Privacy Risks,” was recognized during the 14th International Conference on Computers, Privacy, and Data Protection, European Union’s largest conference in security and privacy. The effort was led by Rakibul Hasan, a research assistant at the Luddy School, who collaborated with Associate Professor of Informatics and Computing David Crandall and Associate Professor of Computer Science Apu Kapadia at Luddy, and Professor Mario Fritz from the CISPA Helmholtz Center for Information Security in Germany.
“I presented the idea during a brain-storming session involving our privacy lab,” Hasan said. “It always feels good to be recognized. But I also take it as a sign that I am following the right direction to solve the problems that I care about, which, in turn, boosts my confidence and inspires me to carry on.”
Researchers cataloged the rationale humans use to distinguish between subjects and bystanders in an image, and systematically validated a set of intuitive concepts that can be used to automatically identify bystanders. Then, they extracted and mapped image features to these concepts and trained models to detect bystanders in photos so that their privacy can be protected by obfuscating them before posting the photos on social media.
“Our model makes this distinction solely based on the visual data in the image using computer vision and machine learning,” Hasan said. “It can be easily integrated into multiple domains including social media and photo-taking devices, such as traditional cameras, mobile phones, and wearable cameras, so that whenever photos are taken or uploaded, bystanders' privacy can be protected, making our solution practical and scalable.”
The CNIL-Inria Privacy Protection Prize jury highlighted the study for having exceptional research in privacy protection with a very high impact on both society and industry. Hasan hopes the award will lead to future collaborations and external funding for further research.
“We’re extremely proud of the work being conducted by our researchers at Luddy, and Rakibul and his group are richly deserving of this recognition,” said Kay Connelly, the associate dean for research at the Luddy School. “Collaborations such as this create so many opportunities for innovation, and this effort at protecting privacy addresses a critical issue for so many people.”