Off the coast of Maine, crawling along the floor of the Atlantic Ocean, lives one of the state’s most culturally and economically significant critters: the lobster. This small, clawed crustacean is responsible for a massive fishing industry, a surfeit of seafood restaurants, and any number of kitschy Vacationland souvenirs. But the industries that revolve around lobster fishing have also generated an enormous amount of trash, and a substantial amount of that trash has come to rest on the shores of Allen Island, which Colby has owned since 2022.
To study marine debris like the refuse of lobster fishing, Professor Whitney King, the Dr. Frank and Theodora Miselis Professor of Chemistry at Colby, and Dr. Nick Record of the Bigelow Lab for Ocean Sciences, decided to work together. King and Record picked Allen Island as a primary site to conduct their research and collect data.
“They have been monitoring how trash moves in coastal environments, where [trash items] end up, and how human activities contribute to this accumulation,” explained Dr. Tahiya Chowdhury, a postdoctoral fellow at the Davis Institute for Artificial Intelligence who is involved in the two scientists’ work.
At first, their research consisted of manual auditing. Student groups would make the trip to Allen Island, walk around, and note marine debris they discovered. But there was a challenge with the volunteer system: the trash on the island is mostly fishing-related, and as such, it is considered private property.
“That is why the people who are involved in the volunteer process can audit [fishing objects], report them, but they cannot touch them,” explained Chowdhury. Volunteers instead have to communicate the location of the trash, which leads to a gear grab during which lobster fishers reclaim their traps, buoys, and more.
Chowdhury described how the research team later decided to capture aerial footage of the island in order to look for debris. They used one of the College’s two drones to take the images. This data collection process is called remote sensing, and it enables researchers to study a location without having to be physically present. An additional benefit of this technique is its efficiency: it takes much less time to photograph an island with a drone than it does to survey it manually.
“It takes multiple days to cover the entire island with a group of volunteers,” said Chowdhury. “But you can cover the entire island [in] just one hour with a drone.”
After obtaining aerial footage, the researchers had to locate the trash within the photographs. “This is where I came in,” said Chowdhury. She and a Colby student, Ray Wang ’26, started to work on a computational solution to the problem.
“We built a pipeline using different existing techniques from computer vision, like, ‘Given an image, can we detect if there is any trash in it?’ Then, if there is trash, what type of trash it is,” said Chowdhury.
Because their solution revolved around marine debris cleanup, Chowdhury and Wang also had to account for the size and material of the debris. For example, they once found an entire refrigerator on the island, which would be impossible for a single person to lift.
Additionally, Chowdhury and Wang built a mapping system that could visually demonstrate the location of the various debris items across the island. According to Chowdhury, they eventually hope to develop a mobile app that visitors will be able to download in order to locate extant trash and log items that the drone and AI object recognition software missed.
Throughout their work, Chowdhury and Wang have dealt with numerous obstacles. First, developing their AI model’s accuracy was challenging. For example, it once misidentified an orange lobster buoy as a carrot.
Second, weather patterns and fluctuating sea currents make it hard to know with certainty where on the island trash will be, as well as whether it will stay put. On at least one occasion, the marine debris moved after data collection took place.
The director of the Davis Institute for Artificial Intelligence, Professor Amanda Stent, described how this is both a help and a hindrance to the work that the larger team is doing on Allen Island: “When the trash moved, that was frustrating for the AI solution, but it’s great for [King] and people who study climate change and ocean rise, because that gives them an opportunity to look at how water moves and what happens when the tides become high. So every problem is also an opportunity for somebody.”
While the work is still in the research phase, it is progressing: the team recently submitted a paper on their Allen Island efforts. Furthermore, because of the involvement that Colby students have had in the project, it serves as a hands-on, in-field learning experience.
Stent spoke to the educational value of these kinds of projects, especially for students at a liberal arts college: “It’s that situated approach that’s also in the context of a particular other subject area—because AI should always be in service of something—with consideration of the ethical and societal implications… that makes learning AI at a liberal arts college like Colby unique.”
Perhaps the most impressive aspect of the Davis Institute for Artificial Intelligence is that its AI projects are not confined to Allen Island. Instead, the institute is working to integrate ethical AI learning into its many projects and classes, and into a Colby education at large.
Ray Wang, who is involved in the Allen Island project, said of the Davis Institute, “There are actually a good amount of projects going, and I think they are always looking for more people and talent to help work for them.” He encouraged students interested in doing similar work to reach out to members of the Institute and pursue their interests.
~ Elias Kemp `27
Be First to Comment