Recent Rutgers University–Newark Ph.D. alum Will Kuhn has been awarded the prestigious Post-Doctoral Research Fellowship in Biology from the National Science Foundation (NSF).
Kuhn, who has a background in entomology and systematic biology, is using his two-year, $138,000 grant to work with a computer specialist at the University of Tennessee, in Knoxville, to develop software apps and online platforms that will automatically identify dragonfly species using face-detection and other computational methods.
Kuhn is the second RU-N Ph.D. student to win the prestigious NSF post-doctoral fellowship this year. The other was Dominic Evangelista, who won in February. Both worked alongside Associate Professor Jessica Ware, a multiple-award-winning scientist who specializes in evolutionary biology.
Kuhn, who arrived at RU-N in 2011 and finished his Ph.D. this summer, is starting work on his post-doc project this fall.
“I am very grateful to the National Science Foundation for this opportunity and to Jessica for all her support over the years,” says Kuhn. “This Post-doctoral Research Fellowship in Biology is very competitive, and it allows me the chance to be the principal investigator of my project, which is a big step forward for my career.”
Kuhn’s post-doc project evolved out of work he did in Ware’s lab while completing his Ph.D. There, Kuhn developed an automated ID system that was moderately successful: He took computer scans of dragonfly and damselfly wings and compared them with a database of other known species, achieving 90% accuracy for the 30-plus species on which he trained it.
But he found cutting the wings off specimens and having to scan them to be less than practical. He wanted to simply snap a photo in the field, submit it to the system and get an ID. This requires complicated computer modeling, however.
That’s because in scanning wings, they effectively become two-dimensional, and lighting is consistent between images, making them less complicated to work with. Field photos, on the other hand, vary in lighting, background, focus, specimen condition, and the 3D position of the specimen in the photograph, which increases the computational demands, he says.
“This post-doc project is a lot more computer science than evolutionary biology and taxonomy, which I've been trained in, and so I'm going to need a lot more help with the computational methods such as computer vision and machine learning,” says Khun.
To that end, Kuhn will work with Dr. Mongi Abidi, of the University of Tenessee, Knoxville’s Electrical Engineering and Computer Science department. Abidi has been director of the university’s Imaging, Robotics, and Intelligent Systems Laboratory for 28 years, conducting research in the field of three-dimensional imaging and data visualization, as well as robotics and enhancement of medical imaging.
“Dr. Abidi will be a great resource for the computational aspects of my post-doc project. He has so much experience in this field,” says Kuhn. “I’m really looking forward to learning all I can from him and his team and am grateful to be working alongside them.”
The potential impact of Kuhn’s work is significant.
Taxonomy is an often overlooked but fundamental subspecialty of biology. Almost every facet of our lives—from the food we eat to the medicine we take, the houses we live in to the natural scenery we admire—has a connection to biological species on earth, according to Kuhn. But the field is suffering due to funding cuts in recent decades, while the need for taxonomy has never been greater as human activities cause unprecedented species decline.
Kuhn’s project will help to address this by introducing a set of apps and open-source web tools for IDing species from images and for measuring specimen images for comparative studies. These tools will be available on the iPlantCollaborative website. In addition, Kuhn will launch ODOMATIC, a system for identifying dragonflies and damselflies, on the OdonataCentral site, allowing researchers and enthusiasts to accurately identify these insects from images of their wings.
As part of Kuhn’s NSF funding, he’ll also hold a series of free workshops in urban communities in New Jersey and Alabama to increase the diversity of OdonataCentral’s user-base and encourage participation in STEM fields. He’ll also host Google Hangout events to spread the word about ODOMATIC internationally, and he’ll recruit RU-N undergraduates to help design and train ODOMATIC, giving them valuable experience in research, programming and taxonomy.
Ware, who has charted Kuhn’s growth throughout his years at RU-N, is thrilled that his project received NSF funding. She says Kuhn was “a breath of fresh air” when he arrived at RU-N, confident, enthusiastic and brimming with ideas.
“Will then came up with his own remarkable project: automatic landmarking software, which became the focus of his postdoctoral research,” says Ware. “I was overjoyed when he found out NSF would fund his work. Post-docs are hard to come by, and NSF postdocs are even harder. Will’s project is promising and will undoubtedly be transformative.”
PHOTO: Associate Professor Jessica Ware with former Ph.D. student Will Kuhn
(photo by Lawrence Lerner)
