Researchers Utilize AI to Improve Group Photos
ALBANY, N.Y. (Oct. 17, 2024) — Artificial Intelligence as a tool can be used in any number of ways, from the intuitive recommendations of chatbots and smart assistants to autonomous vehicles.
In UAlbany Associate Professor of Computer Science Pradeep Atrey's lab, he and doctoral student Omkar Kulkarni are utilizing AI to create a virtual photo editing tool. Combining techniques of computer vision with machine learning, Atrey and Kulkarni have developed an innovative iOS application designed to improve the quality of group portraits.
Named 'GARGI,' which stands for Gaze-Aware Representative Group Image, the app focuses on addressing the issue of what happens when multiple people in a group photo have different points of focus.
"During social gatherings, multiple group photos are taken, but if the subjects are not looking in the same direction, the image is usually discarded due to its lack of aesthetic quality," said Kulkarni. "This simple yet interesting observation led us to develop an app that enhances the visual appeal of group photos by ensuring gaze uniformity to make the group image more pleasing."
While technologies like Burst mode and Apple Inc.'s Live Photos have been developed to enhance features and provide options for users, scrolling through a vast selection of similar images can be a challenge. This is where GARGI can provide relief.
"Apple's proprietary algorithm overlooks gaze uniformity when selecting a key frame from group photos captured in Live mode. GARGI addresses this gap by identifying a representative image with minimal gaze deviation, resulting in a more aesthetically pleasing outcome compared to traditional methods," said Atrey, who serves as the co-director of the Albany Lab for Privacy and Security at the College of Nanotechnology, Science, and Engineering.
A Fulfilling Academic Journey
For Kulkarni, the project has been particularly fascinating because it combines computer vision with machine learning.
"The challenge of developing algorithms that can detect and select a better representative image in group images has been both technically demanding and rewarding for me," said Kulkarni.
Originally from the small town of Kolhapur in the Maharashtra state of India, Kulkarni cites his parents' unwavering support in his journey to pursue graduate degrees at UAlbany.
"They always emphasized the importance of education and encouraged me to pursue my dreams," said Kulkarni, who has nearly completed his PhD. "This strong foundation motivated me to excel academically and eventually led me to choose UAlbany to pursue my master's and doctoral degrees."
He credits Atrey and his professors at CNSE for helping him to develop strong research skills which have enabled him to formulate, present and solve complex problems in his field.
"My work on the GARGI app and my overall research have deepened my technical abilities in computer vision, machine learning and AI," said Kulkarni. "I have also relished the opportunity to work in the ALPS lab with Dr. Atrey and my fellow researchers, which provided a great opportunity to collaborate with diverse individuals and helped me to understand various cultures, both in education and life."
"When I discussed the everyday real-world problem of gaze uniformity with Omkar, he immediately recognized its potential and chose to focus on it for his PhD research," said Atrey. "Although it may seem trivial, solving the issue of gaze uniformity required rigorous scientific effort, which Omkar led exceptionally well."
Celebrating Different Cultures
Apart from his studies, Kulkarni has been actively involved with the Indian Student Organization at UAlbany.
"I began as a volunteer in 2018 and eventually served as president for two consecutive terms in 2019 and 2020. The organization is renowned at UAlbany for its grand ‘Diwali Night’ celebration, which honors the Indian festival of lights, along with various other activities on campus," said Kulkarni. "This event allowed me to reconnect with my cultural roots while sharing our traditions with the broader university community."
In 2019, under his leadership, the Indian Student Organization hosted the largest Diwali Night celebration in the University's history, earning an award from UAlbany's Student Organization Resource Center.
"This experience was incredibly rewarding and memorable, as it not only celebrated our heritage but also fostered a sense of community and inclusivity," said Kulkarni.
After completing his studies, Kulkarni hopes to continue to pursue his research, focusing on solving real-world problems through technology.
"I plan to pursue a career in computer vision, aiming for positions that emphasize research and development for the greater good of society," said Kulkarni. "Additionally, I have a passion for teaching and aspire to guide students in India to pursue their dreams and goals globally."
AI Research and Collaboration
The GARGI app served as a key component of Kulkarni’s PhD dissertation, under the direction of Atrey and Rutgers University Associate Professor of Library and Information Science Vivek Singh.
Significant contributions to the app’s development were also made by UAlbany Class of 2024 graduates Thomas Lloyd-Jones of Albany, My Tran of Queensbury and Gregory Vincent of Schenectady.
"Thomas, My and Greg eagerly joined the team and did an excellent job of implementing the gaze uniformity algorithm in the GARGI app," said Atrey. "They excelled in their work and were an integral part of the project."
Next week, AI will be front and center during UAlbany's second annual Research and Entrepreneurship Week. Among the four-day block, Day 2 will focus on how AI is transforming the local and global landscape on Oct. 22 during "Capital Region Artificial Intelligence Day."
Overall, UAlbany continues to expand its academic and research footprint through the AI Plus Initiative. This includes first-in-the-nation supercomputing collaborations with NVIDIA to adopt the NVIDIA DGX Cloud and with IBM to install the first prototype IBM Artificial Intelligence Unit (IBM AIU) at a university to advance generative AI research.
On Oct. 10, UAlbany President Havidán Rodríguez was joined by New York State Gov. Kathy Hochul and SUNY Chancellor John B. King Jr. to unveil one of the most advanced AI supercomputers at a U.S. university.
The new $16.5M supercomputer, powered by NVIDIA DGX systems, is currently the most capable AI supercomputer within SUNY and a central pillar of AI Plus.