AI drones track world's rarest dolphins from extinction
November 03, 2025 · 3 min read
Off the west coast of New Zealand's North Island, a technological rescue mission is underway for one of the planet's most endangered marine mammals. The Maui dolphin, a subspecies with distinctive rounded dorsal fins and a known population of just 54 individuals, faces extinction after decades of fishing practices pushed it to the brink.
Now, conservation group MAUI63 is deploying AI-equipped drones to study these elusive creatures in ways previously impossible. The non-profit organization, formed in 2018 when the population stood at 63 dolphins, combines marine biology expertise with cutting-edge technology to address critical knowledge gaps about the species' behavior and habitat.
According to co-founder Professor Rochelle Constantine, a marine biologist, current understanding of Maui dolphins comes almost entirely from summer observations. "We know virtually nothing about them in winter," she says, noting that deteriorating weather conditions have traditionally made monitoring difficult and costly. The AI-powered solution aims to provide year-round data collection.
The technology developed by MAUI63 represents a significant engineering challenge. CEO Tane van der Boon spent months teaching himself to build custom computer vision models after discovering existing algorithms couldn't identify Maui dolphins' unique fin shapes. He manually tagged dolphin images from internet footage to train the system, beginning a four-year development process.
The current system combines an 8K ultra high-definition camera with object detection models and open-source algorithms originally developed for facial recognition. Hosted on Microsoft Azure, the platform gathers data to identify individual dolphins by analyzing dorsal fin shapes, sizes, and distinctive markings.
Funding from New Zealand's Cloud and AI Country plan and Microsoft Philanthropies ANZ has supported the project's development. The team also had to obtain specialized qualifications to operate their 4.5-meter wingspan drone over open ocean, achieving their first successful dolphin detection earlier this year.
Beyond basic monitoring, MAUI63 is developing a citizen science app called Sea Spotter that will allow public sightings to contribute to the database. The organization is also working on real-time alert systems for fishing companies, potentially reducing bycatch risks when dolphins stray outside protected marine sanctuaries.
Van der Boon emphasizes that MAUI63's role is data collection rather than policy advocacy. "We're just trying to collect the data and make it available to anyone who needs it," he says, acknowledging that while success isn't guaranteed, the effort represents a crucial attempt to prevent extinction.
The technology may soon benefit other endangered species, with potential applications in Antarctica being explored with the European Union Environmental Council. For Constantine, the high-tech approach has already transformed conservation conversations, providing governments and the public with better information for making critical protection decisions.