A pair of puffins on the Latrabjarg bird cliffs on the northwestern tip of Iceland.

AI helps to detect and count puffins

Image credit: Anthony Hathaway/Dreamstime

Wind farm developer SSE Renewables is trialling a new technique to count puffins using artificial intelligence (AI) and image-recognition technology.

The project, which is supported by Microsoft, Avanade and nature reserve NatureScot, could transform the way animal colonies are counted when companies such as those building developments want to understand the impact on local wildlife, the operator said. The trials are taking place on the Isle of May off the coast of Scotland.

In order to count puffins, traditionally rangers would lie on the ground and put their hand into burrows to feel for a pair of puffins and their egg. But very often, the birds would give these rangers a bite or a scratch. The rangers would check tens of thousands in an area.

While they are not at imminent risk of extinction, puffins are on the Birds of Conservation Concern 4 Red List, meaning that there are serious concerns over numbers in the wild. One reason for this is puffin couples only lay one egg a year. It’s why rangers take part in counting projects in places such as the Farne Islands, the Shetland Islands and the Isle of May, to understand whether the population is growing or shrinking.

According to SSE Renewables, the work being done on the Isle of May to spot, recognise and count puffins using technology could help minimise disruption to birds’ breeding and feeding habits as those sustainability projects move forward.

“As the investment in renewable energy continues, it’s even more important to ensure that developments, like wind farms, are not having any detrimental environmental effects,” said Simon Turner, CTO of Data and AI at Avanade.

“This area on the Isle of May attracts many puffins each year to breed, and it’s key to ensure that with the planning of any new wind farms this is not interrupted.”

To monitor the puffin population on the island, Turner said NatureScot would normally send people with clipboards to sit for hours, marking down how many puffins they saw. “With SSE Renewables, we saw an opportunity to use technology to make this process more accurate, more efficient, and less invasive for the puffins. Using cameras and AI, we can now count the number of puffins and monitor their burrows all day, every day, without going near them.”

SSE Renewables and Avanade have placed four cameras in stainless steel boxes on the island to capture live footage of the puffins after they begin their return to land to breed in late March/early April following eight months at sea. They will make the island their home for the spring and summer months – using the same burrows as previous years – before returning to the sea in mid-August.

Each box has a condensation heater, wipers to cope with the miserable, or 'dreich', weather, and a backup power supply. The data captured by the cameras is stored in a Microsoft Azure Data Lake and uses the Azure Kubernetes Service, which has the power and elasticity to handle vast amounts of information.

Clare Barclay, chief executive officer at Microsoft UK, said: “The innovative puffin-monitoring project on the Isle of May shows the impact technology can have on advancing sustainability and is just one initial example of how we are collaborating with SSE to shape a more sustainable future.”

Avanade and SSE Renewables trained the AI using an “image recognition” model, which is fed simple pictures until it learns to certify structures such as lines, squares, and circles. Subsequent images become more complex, allowing the AI model to recognise objects. This model was fed images of puffins and other random images and told which were puffins and which were not. When the cameras on the Isle of May are turned on, the AI can spot the puffins, separate them from background images such as rocks, and track them, frame by frame, as they move around.

“The AI will draw a box around each puffin it spots and give them unique tags like ‘001, 002, 003’ etc,” Turner added. “When the camera moves to the next frame, the AI understands that the puffin closest to a particular box is the same puffin. It’s just taken a step to the left or right, so it redraws the box around the bird. This happens repeatedly for every frame of the footage. Even if the puffin flies out of the frame, the AI system will recognise where it flew out of sight, and attempt to track it again if it comes back. That’s how we can track and count individual puffins.”

Puffins are mostly nocturnal when they nest on land, so the researchers had to adjust the camera’s brightness and resolution settings to capture footage of the birds. They couldn’t use night vision, as the AI model had been trained on images of puffins taken in daylight.

SSE Renewables is also keen to find out if its Beatrice wind farm is affecting the birds’ flight paths as they travel to gather food to take back to their burrows.

Beatrice, located 13km off the Caithness coast, comprises 84 turbines, each with a 154m diameter rotor. A joint venture between SSE Renewables, Red Rock Power, the Renewables Infrastructure Group, and Equitex, Beatrice became operational in 2019 and can produce 588MW of energy – enough to power up to 450,000 homes. The electricity generated by the turbines travels along sub-sea and underground cables before coming ashore near Keith in Moray and connecting to the Blackhillock substation.

“We wanted to improve our environmental monitoring so we can be more proactive in that area. It’s part of a digital transformation at SSE Renewables that Microsoft and Avanade have been helping us with,” said Oliver Abell, account manager at SSE Renewables’ Engineering Centre. “There’s no way SSE Renewables could have started the AI puffin-counting project on our own; we’re an energy company, not a technology company. Microsoft and Avanade have been critical pieces of this project.”

While the Isle of May project is focused on puffins, there are hopes it could monitor other animals who may be affected by developments “We could monitor salmon to make sure they can migrate in rivers, for example,” Abell added. “This technology could work in any environment in which you want to monitor a species and be hands-off, because it’s too remote or because you don’t want humans interfering in that environment.”

Having hatched and raised their pufflings (baby puffins), the puffins have now left the Isle of May for another season. SSE Renewables, Microsoft, and Avanade will now look at the data to learn if their AI project has been a success.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles