The Data Lab’s data scientists in collaboration with Heriot-Watt University, Orkney Sustainable Fisheries, and Bangor University have unlocked the key to accurately measuring lobster stock like never before.
By melding low-cost cameras with the magic of machine learning, this ground breaking solution will help fishers, researchers, and sustainable fisheries management by reducing costs and carbon footprint and improving environmental impact through increased efficiency.
To manage our fisheries sustainably, we must base our decisions on solid evidence. One crucial piece of information is the catch rates in commercial fishing, which tell us how many fish are being caught.
However, there’s a catch (no pun intended!) – these catch rates can be affected by factors other than the number of fish in the area.
For instance, in creel (trap) fisheries, how fish are caught is influenced by their behaviour and interactions with other species. This makes it challenging to determine the number of fish accurately. And when it comes to monitoring European lobsters, things get even trickier!
Their cryptic behaviour (meaning they’re good at hiding) makes it tough to observe and sample them in a quantifiable manner, especially when compared to species like the American lobster.
Building on the promising results from the initial feasibility study, a full-scale Research and Development (R&D) project was launched. The primary goal of this was to create an accessible and effective tool that could revolutionise the way we manage fish populations in local inshore fisheries.
The approach involved developing a low-cost creel-mounted underwater camera system carefully designed to cater to fishers, scientists, and managers’ needs. The aim was to better understand fish behaviour and interactions in real-world fishing scenarios.
We integrated machine learning into the project to add an extra layer of cutting-edge technology. The vision was to leverage machine learning algorithms to detect and quantify the presence of European lobsters near fishing activity.
This unprecedented combination of systems and technologies in that setting had never been attempted. Initially designed for European lobsters, the method is being extended to crab stock monitoring, unlocking new possibilities for preserving diverse marine life.
The Data Lab’s talented data scientists played a crucial role in this project. They undertook the essential task of pre-processing the vast number of videos and images captured by the underwater camera system. They prepared the groundwork for training a deep learning method through meticulous data handling. This advanced technique automatically recognises and detects the elusive lobsters captured on camera, helping us gather vital information to support sustainable management practices.
Our solution aimed to bridge the gap between the complexities of fish behaviour and the need for accurate data. By embracing innovation and collaboration, we aspired to create a game-changing tool to ensure the longevity of our precious marine ecosystems and safeguard the livelihoods of coastal communities.
The impact of this breakthrough is already being felt in Orkney, where the technology has been integrated into the practices of Orkney Sustainable Fisheries. Marine Scotland Science has also embraced this solution and is using it to support critical consultations on inshore fisheries management in Scotland.
Wider uptake within the UK and beyond is anticipated, with the potential for adopting this technology in the Irish Sea, where Bangor University is collaborating with the Isle of Man government.
As we look towards the long-term impact of our efforts, we envision substantial benefits to the seafood sector. By enabling fishers to target lobster stocks more efficiently, we anticipate reduced costs, a diminished carbon footprint, and an overall improvement in environmental impact. This solution seeks to strike the delicate balance between thriving fisheries and safeguarding the health of our oceans.