Six seals, two researchers and a secret Finn

Sometimes two days in an aquarium can provide more results than months of fieldwork outside. The seals at the Polaria aquarium helped teach computers and machine‑vision algorithms to automatically identify seals. Our goal is to develop tools that allow us to better reconcile seals and fisheries in the future.

In February, we spent two intensive days at the Polaria aquarium in Tromsø, Norway. The visit was part of research projects investigating the impacts of seals on fisheries and exploring ways to balance growing seal populations with the fishing industry.

Why monitor seals with a sonar?

Seals are intelligent and adaptable animals. They quickly learn to take advantage of the easy feeding opportunities provided by fishing and fish farming. This causes economic losses to fisheries and also poses risks to seals themselves, which can become entangled in fishing gear.

In our projects, we have developed automatic tools based on fishing sonars to count and monitor seals. The goal is to teach a computer to recognize a seal as it swims past the sonar. Once this works, we can automatically collect data around fishing gear and aquaculture facilities. Accurate information helps us understand how and when seals move near traps and net pens. That knowledge can guide effective solutions that protect both the gear and the seals.

The aquarium provided a unique research environment

We had previously attempted similar data collection in the field. However, over a couple of months we only had a handful of observations. At Polaria, the situation was completely different.

The aquarium houses seven seals: bearded seals (Erignathus barbatus), harbour seals (Phoca vitulina), and a ringed seal (Pusa hispida). Ringed seals also occur in Finland, and the others are similar in size to grey seals. The seals moved actively in the pool almost nonstop. At most, only a few seconds passed without a seal appearing in the sonar image.

We were able to record plenty of video showing the seals’ underwater movements from the sonar’s perspective. This is exactly the kind of material needed to train a machine‑vision algorithm. Polaria also deserves special thanks for excellent cooperation: everything ran smoothly, and the staff did everything possible to support the project.

Curious seals and one sulking ringed seal

Working close to the seals was an experience in itself. Six of the seven seals constantly lingered near the sonar. They did not seem disturbed by the device; quite the opposite they were curious to see what we were doing.

The lone ringed seal, Junior, reacted differently. He stayed in the outdoor pool and observed the indoor activities from a distance. Although Junior was born in the Netherlands and has also lived in France, his desire for solitude made us wonder whether he might have some Finnish heritage. Even though he seemed content, we were left wondering whether we should send Junior a sauna to use by the outdoor pool.