Drones and AI give bird’s-eye view of seabird colonies

Scientists at Duke University and the Wildlife Conservation Society (WCS) used a deep-learning algorithm – a form of AI – to analyse over 10,000 drone images of mixed colonies of seabirds in the Malvinas/Falkland Islands off Argentina’s coast

The Malvinas/Falklands are home to the world’s largest colonies of black-browed albatrosses and second-largest colonies of southern rockhopper penguins. Hundreds of thousands of birds breed on the islands in densely interspersed groups.

According to the researchers, the deep-learning algorithm correctly identified and counted the albatrosses with 97 per cent accuracy and the penguins with 87 per cent. The automated counts were within 5 per cent of human counts, about 90 per cent of the time. 

“Using drone surveys and deep learning gives us an alternative that is remarkably accurate, less disruptive and significantly easier,” said Madeline C Hayes, a remote-sensing analyst at the Duke University Marine Lab. “One person, or a small team, can do it, and the equipment you need to do it isn’t all that costly or complicated.”

Previously, teams of scientists have been monitoring the colonies, which are located on two rocky, uninhabited outer islands. They count the number of each species they observe on a portion of the islands and extrapolate those numbers to get population estimates for the full colonies.

Because the colonies are large and densely interspersed and the penguins are much smaller than the albatrosses (and, thus, easy to miss), counts often need to be repeated. The presence of the scientists can also disrupt the birds’ breeding and nursing behaviours.

To conduct the new surveys, WCS scientists used an off-the-shelf consumer drone to collect over 10,000 individual photos. Hayes then converted these into a large-scale composite visual using image-processing software.

She then analysed the image using a convolutional neural network (CNN), a type of AI that employs a deep-learning algorithm to analyse an image and differentiate and count the objects it 'sees' in it – in this case, two different species of sea birds. The team added together these counts to create comprehensive estimates of the total number of birds found in the colonies.

“We loosely modelled a CNN on the human neural network, where it learns from experience,” said David W Johnston, director of the Duke Marine Robotics and Remote Sensing Lab. “Essentially, the computer identifies different visual patterns, like those made by black-browed albatrosses or southern rockhopper penguins in sample images, and over time it learns how to identify the objects forming those patterns in other images such as our composite photo.”

Johnston, who is also an associate professor of the practice of marine conservation ecology at Duke’s Nicholas School of the Environment, said the emerging drone- and CNN-enabled approach is widely applicable “and increases our ability to monitor the size and health of seabird colonies worldwide, and the health of the marine ecosystems they inhabit”.

Guillermo Harris, a senior conservationist at WCS, added: “Counting large seabird colonies of mixed species at remote locations has been an ongoing challenge for conservationists. This technology will contribute to regular population assessments of some species, helping us better understand whether conservation efforts are working.”

The two islands used for the pilot study, Grand Jason Island and Steeple Jason Island, are important for conservation and a focal area of work by WCS.