Study highlights the potential of AI surveillance at fish cleaning tables to provide inexpensive and non-intrusive data collection
University of Wollongong (UOW) researchers Lachlan Baker and Dr Katharina Peters have co-authored a study exploring the use of artificial intelligence (AI) to monitor recreational fishing activities. Published in the New Zealand Journal of Marine and Freshwater Research, the research presents a fresh approach to data collection, offering significant potential to enhance fisheries management and sustainability.
The study highlights the potential of AI surveillance at fish cleaning tables to provide inexpensive, easily accessible, and non-intrusive data collection. A 24/7 monitoring framework, implemented at boat ramp cleaning tables, can identify fish species and measure individual fish, delivering an unprecedented level of data quantity and quality in an industry that has historically been data-poor. This research can aid in understanding fish stocks, improving recreational fisheries management, and ensuring that resources are exploited responsibly and sustainably.
"Marine recreational fishing is an exceedingly popular activity with immense economic and social value for millions of people around the world," Dr Peters said.
"It's a booming industry which is often overshadowed by the commercial fishing sector, leaving the impacts of recreational fishing relatively under-researched. Traditional monitoring techniques, such as voluntary surveys, are resource-intensive and often lack the precision needed for effective management."
The research involved the development and testing of AI models to identify fish species and measure their size using images captured at cleaning tables. Two models were assessed: a basic image classification model and a more advanced object detection model. The object detection model demonstrated significantly superior performance, achieving 80 per cent accuracy in species identification compared to 30 per cent for the simpler classification model.
"Recreational fishing contributes billions to the economy and plays a vital role in community and environmental engagement. However, monitoring its impact on fish populations has been constrained by outdated methods. Our study demonstrates that AI can revolutionise data collection, offering continuous, cost-effective solutions," Mr Baker said.
Lachlan Baker conducting field tests of the AI identification app at the Bellambi Hotel Fishing Club competition, manipulating camera height, fish orientation, direction, and species to test the AI model. The event formed a large part of the data collection process for the paper.
The study also analysed variables influencing AI accuracy, such as camera height, image resolution, and fish orientation. It found that lower camera heights and dorsal fish orientations provided the best results, while poor image quality reduced measurement accuracy.
Dr Peters said the study could have broader implications on the industry.
"Our findings pave the way for scalable AI systems that can operate 24/7 along coastlines, providing vital data to ensure sustainable fisheries management," Dr Peters said.
"This is a step toward responsible resource use and long-term ecological balance."
This research, supported by the Recreational Fishing Trust Grant by NSW Government, marks a significant step forward in applying cutting-edge AI technology to real-world challenges, enabling more efficient and accurate monitoring of recreational fishing activities.
