Mui dolphins are one of the rarest and most endangered dolphins in the sea. They are small and have a round dorsal fin that makes them stand out. This subspecies is almost extinct because of fishing methods like gillnetting that have been used for decades off the west coast of New Zealand in the South Pacific.
Now, scientists and people who work to protect the environment are using drones, AI, and the cloud to find out more about these rare marine mammals. They say that the solution can also be used to study other species in the oceans that are fighting for their lives.
The project is part of a growing trend to use AI and other technologies to collect and analyse data about the environment in a more effective way. For example, Conservation Metrics, a partner of Microsoft AI for Earth, uses machine learning, remote sensing, and scientific know-how to make wildlife surveys bigger and better. Another partner organisation, NatureServe, uses Esri ArcGIS tools and Microsoft cloud computing to make high-resolution habitat maps for species that are in danger.
Scientists and conservationists with the non-profit group MAUI63 are using artificial intelligence (AI) and other tools to help protect the Mui dolphins, which are named after the Polynesian god Mui.
Mui dolphins are an important part of Aotearoa, which is the Mori name for New Zealand. They are also a spiritual and ecological part of this country. They live in the water off the west coast of the North Island of the country, which is also called Te Ika-a-Mui, which means “the fish of Mui.”
When fully grown, Mui dolphins can be up to 1.7 metres long and weigh up to 50 kilogrammes. They are one of the smallest marine dolphins and also one of the hardest to find. They have white, grey, and black spots, and their back fins are round and black. Markings don’t change from animal to animal like human facial features do, so you can’t tell which animal is which just by looking at it. The old ways of keeping track of and studying these fast-moving animals at sea have been hard and expensive. Researchers say they don’t know much about how they act, especially in the winter when the weather gets bad.
Now, MAUI63 thinks it has a solution: an AI-powered drone that can find, track, and identify dolphins quickly and easily. Professor Rochelle Constantine, who is also one of the founders and a marine biologist, says that the goal of their work is to “give certainty to our uncertainty.”
“Right now, all we know about them comes from the summer. “In the winter, we know almost nothing about them,” she says.
In 2018, Constantine, Tane van der Boon, an expert in technology and new ideas, and Willy Wang, a fan of drones, formed MAUI63. At the time, it was thought that there were about 63 Mui dolphins. Since then, that number has gone down to 54.
Van der Boon, who is the CEO of the group, and Wang came up with the idea of using drones, machine learning, and cloud computing to study dolphins while they were having drinks at a pub. “I was getting interested in how computers learn, and I saw how amazing it is to teach a computer to see. “I was really interested in everything we could start to solve and do,” he says.
The fins of Mui dolphins are round, while the fins of other dolphins are more pointed. That meant that computer vision models that could be used to find Mui dolphins were not good enough. So, van der Boon taught himself how to build a model over “a couple of months of nights and weekends.” Then, he carefully tagged Mui dolphin pictures from the internet to teach it to recognise them.
It was the first of many hard things to do. The next four years were spent developing, testing, and raising money. For the team to fly their 4.5-meter-wide drone out to sea, they also had to get special training. The first Mui dolphins they saw were earlier this year.
“It was a lot of fun. “We were in the van, and the drone was 16 kilometres down the coast. We could see the AI detecting dolphins as we went in circles around them,” van der Boon says.
New Zealand’s Cloud and AI Country Plan, which includes funding for projects with long-term effects on society, and Microsoft Philanthropies ANZ have both helped with development. The solution uses an 8K ultra-high-definition still camera, a full HD gimbal camera, an object detection model for finding dolphins, and an open-source algorithm that was originally made for facial recognition. It is hosted on Microsoft Azure and collects data that will be used to figure out how to identify each animal based on the shape and size of its dorsal fin and any scratches or marks it has.
MAUI63 is also working on an app called Sea Spotter, which is being funded by Microsoft. This app uses Azure Functions to let people upload photos of Mui they’ve seen and use an AI algorithm to figure out who they saw. Conservationists say that it is important to know where the Mui dolphin lives in order to know how to protect them from dangers.
Constantine says that there is now a “extremely low” chance that Mui dolphins will be accidentally caught in fishing nets. This is because a marine sanctuary was set up around their known habitat in 2008 and will be expanded in 2020. Still, they may wander outside of these safe areas. So, MAUI63 is working with fishing companies on a project to connect them to the drone so that the crews can be notified in real time of what the drone sees.
Toxoplasmosis, which is caused by a parasite that lives in cat poop, is another danger. It gets into the marine food chain through runoff from land. This makes marine mammals have stillbirths and die. “If you know where dolphins are often, you can start to look for places where toxoplasmosis might be getting into the water, and maybe something can be done about that,” says van der Boon.
The goal of MAUI63 is to give people who make decisions about conservation information that is backed up by science. “All we want to do is gather the information and make it available to anyone who needs it. We’re not here to say how they should or shouldn’t be kept safe. This is important to us because people have very different ideas about it,” says van der Boon. At this point, he says, it’s not likely that MAUI63’s work will help save the species, but everyone agrees that it’s worth a shot.
Many indigenous Mori people think of Mui dolphins as kaitiaki (guardians) who helped guide their ancestors’ canoes when they first came to Aotearoa (New Zealand) hundreds of years ago.
Dr. Aroha Spinks, an environmental scientist, says that protecting them is important if we want the mauri, or life force, of the environment to grow. “From a Mori point of view, which is also supported by science, the health of the environment affects the health and well-being of the people,” she says.
MAUI63 wants to share what it has learned and the technology it has developed with people who work with other marine species. For example, the European Union Environmental Council and MAUI63 might work together on a project in Antarctica. Constantine hopes that other researchers will find the high-tech method as useful as it has been for her. “It makes a huge difference in my world, the conversations I can have, and the information we can give to governments and the public about how to make conservation decisions that really matter.”
Image on top: MAUI63 uses drones, artificial intelligence, and the cloud to learn more about Maui dolphins. The video was made by MAUI63.
