Finding solutions to improve turtle identification and support machine learning projects across Africa
Protecting the ecosystems around us is vital to safeguarding the future of our planet and all its living citizens. Fortunately, new artificial intelligence (AI) systems are making progress in conservation efforts worldwide, helping to tackle complex problems at scale – from studying the behavior of animal communities in the Serengeti to preserving the declining ecosystem, to spotting poachers and their wounded prey to prevent species extinction.
As part of our mission to help benefit humanity with the technologies we develop, it is important to ensure that diverse groups of people build the artificial intelligence systems of the future so that they are fair and just. This includes expanding the machine learning (ML) community and engaging with a wider audience to tackle important problems using artificial intelligence.
Through research, we came up with Ziddy – a dedicated partner with complementary goals – who are the largest community of African data scientists and host competitions focused on solving Africa’s most pressing problems.
Us Scientific teamHis Diversity, Equity, and Inclusion (DE&I) team worked with Zindi to identify a scientific challenge that could help advance conservation efforts and develop participation in AI. Inspired by Zindi turtle delineation challengewe landed on a project with the potential for real impact: turtle facial recognition.
Biologists consider turtles as an indicator species. These are classes of organisms whose behavior helps scientists understand the underlying well-being of their ecosystem. For example, the presence of otters in rivers has been considered a sign of a clean, healthy river since the banning of chlorine pesticides in the 1970s brought the species back from the brink of extinction.
Turtles are another such species. Grazing on sea cover, they nurture the ecosystem, providing habitat for many fish and crustaceans. Traditionally, individual turtles have been identified and tracked by biologists with physical tags, although the frequent loss or erosion of these tags in seawater has made this method unreliable. To help solve some of these challenges, we launched an ML challenge called Turtle recall.
Given the added challenge of holding a turtle still enough to locate its tag, the turtle recall challenge was intended to circumvent these problems with turtle face recognition. This is possible because the pattern of scales on a turtle’s face is unique to the individual and remains the same over a lifetime of many decades.
The challenge aimed to increase the reliability and speed of turtle identification and potentially provide a way to completely replace the use of inconvenient physical tags. To make this possible, we needed a data set to work with. Fortunately, after Zindi’s previous turtle-based challenge with a Kenya-based charity Local Ocean Protectionteams were kindly able to share a dataset of turtle face labels.
The competition started in November 2021 and lasted for five months. To encourage competitor participation, the team implemented a notebook colaban in-browser programming environment, which introduced two common programming tools: JAX and Haiku.
Participants were tasked with downloading the challenge data and training models to predict the identity of a turtle as accurately as possible given a photo taken from a specific angle. Having submitted their predictions on the data hidden from the model, they were able to visit a public leaderboard tracking each participant’s progress.
Community engagement has been incredibly positive, as has the technical innovation demonstrated by the teams during the challenge. During the competition, we received submissions from a wide range of AI enthusiasts from 13 different African countries – including countries not traditionally well represented at the biggest ML conferences, such as Ghana and Benin.
Our turtle conservation partners have indicated that the participant’s level of prediction accuracy will be immediately useful for locating turtles in the field, meaning these models can have a real and immediate impact on wildlife conservation.
As part of Zindi’s ongoing efforts to support climate positive challenges, they are also working Swahili sound classification in Kenya to assist translation and emergency services, and air quality prediction in Uganda to improve social welfare.
We are grateful to Zindi for their collaboration and to all those who have contributed their time to the Turtle Recall challenge and the growing field of artificial intelligence for conservation. And we look forward to seeing how people around the world continue to find ways to apply AI technologies to build a healthy, sustainable future for the planet.
Read more about Turtle Recall at Zindi’s blog and learn about Zindi at