The future of ecological conservation lies in innovative technologies that streamline data collection and analysis. One such revolution in conservation science is integrating drones with artificial intelligence (AI) for wildlife monitoring. This offers a non-invasive, efficient, and cost-effective way to gather information, but unfortunately it’s not always straightforward! We’ve recently been working with Olivier Decitre for his Masters research project at James Cook University to address this challenge. He wanted to develop a workflow using drone mapping, AI, and geospatial information systems (GIS), to detect and map giant clams on the Great Barrier Reef. So this case study is his story.
While you can read the full published paper here, in this blog post we’ll cover the key points of the work. Our case study offers insights into the broader applications of these technologies for environmental conservation.
Banner image by NOAA via Unsplash.
Contents
Why monitor giant clams?
Giant clams (Tridacninae) play a crucial role in the health of reef ecosystems. Not only do they contribute to reef structure by providing carbonate material, but they also act as biofilters, preventing algal blooms that can damage coral reefs. Unfortunately, giant clams are increasingly threatened by overfishing, habitat destruction, and climate change. Thus, effective monitoring is critical to conserving these species and maintaining the health of coral reef ecosystems.
Traditionally, giant clam monitoring has been labor-intensive and expensive. Divers have had to physically count clams along transects, covering only small areas of the reef. However, drone technology offers a new solution, enabling researchers to collect data over large areas more quickly and efficiently.
Drone mapping + AI for wildlife monitoring
Drone mapping has revolutionized how we monitor remote and expansive ecosystems like the Great Barrier Reef. However, the sheer volume of data that drones generate can overwhelm manual data processing. This is where AI, specifically object detection models, come into play.
As the giant clam can grow to over one meter in size, under the right conditions we can easily see them in drone mapping data. It gets a bit more challenging if the water is deep or murky, but luckily we have a fantastic study site to get started.
And Pioneer Bay at Goolboddi (Orpheus) Island is exceptional! Here, there are many clams in high density and a variety of colours. We also have many drone mapping datasets available on GeoNadir with differing water and lighting conditions that help make our AI model more robust.
Once we had our data, Olivier worked with YOLOv5, a state-of-the-art AI object detection model, to identify and count giant clams across large areas of the reef. He was even able to apply it to unseen data on other reefs with success.
How it works
Key findings and research benefits
Efficiency and scalability: By automating the detection process, we can analyze much larger areas than would be possible with traditional methods. The model detected up to seven clams per square meter in some parts of the reef!
Accuracy and adaptability: The YOLOv5 model achieved high accuracy in detecting clams, even under varying environmental conditions. It also demonstrated adaptability, successfully identifying clams in unseen test locations.
Cost-effectiveness: Once the model is trained, it can be applied repeatedly to new datasets, making it a cost-effective tool for long-term monitoring. This reduces the need for costly and time-consuming field surveys.
Conservation applications: The ability to detect and map clams at this scale provides valuable insights for conservationists and fisheries managers. By understanding where clams are concentrated, they can focus their efforts on protecting the most critical habitats and monitor population trends over time.
Beyond giant clams: broader applications of AI for wildlife monitoring
While this research focused on giant clams, these methods can easily be adapted to other species and ecosystems. From monitoring seabird colonies (Olivier has tested this already!) to tracking coral health, combining drone mapping and AI has broad applications for wildlife conservation. As more datasets become available, these models can be trained to detect a wide range of species, offering a powerful tool for global conservation efforts.
The open-access nature of GeoNadir’s repository means that researchers from anywhere can contribute to, or use the data. This collaborative approach accelerates AI model development like the one used in this study, as more diverse datasets lead to more robust and adaptable models.
A bright future for conservation tech
As drone technology and AI continue to advance, the possibilities for wildlife monitoring are expanding rapidly. GeoNadir’s research on the Great Barrier Reef demonstrates how these technologies can revolutionize conservation efforts by offering a scalable, cost-effective, and accurate way to monitor endangered species like giant clams. The workflow developed here is a blueprint for future ecological monitoring, providing conservationists with the tools they need to protect vulnerable ecosystems.
By embracing the power of drone mapping and AI, we can ensure a brighter future for our planet’s wildlife and ecosystems. If you’re interested in learning more about how drone data can transform conservation, visit GeoNadir’s website and explore the wealth of resources available to researchers and conservationists alike.
