Machine learning methods for fishing gears identification from GPS trajectory data: application to inter-regional fisheries
Year: 2022
Type: preprint
Abstract: A key challenge in marine ecosystems management and protection is to identify and evaluate the spatial distribution of human activities, such as fishing, and their environmental impact. This involve gathering information on the spatiotemporal distribution of fishing vessels and the use of their related gears. In this context, recent technological advances in the collection of geolocation data and associated legislation have made it possible to collect information on the fleets of artisanal fishing vessels. These data were first collected for safety and to closely monitor fisheries, but the recent development of machine learning methods have greatly increased the potential use of these data. Here we present an approach to identify fishing gear exploiting both machine learning and vessel behavior analysis. Using a large dataset containing 25612 vessel tracks gathered from 2006 to 2020 in the European waters, we developed a method to detect 10 fishing gears classes by analyzing vessel behavior only. The classification accuracy using this approach is 94.89 % demonstrating the relevance of the proposed approach for monitoring fisheries in European waters. This method only exploiting motion information independently of the fishing location makes the demonstration of a versatile approach of machine learning to detect fishing gear at an inter-regional scale. It may open new avenues to operational monitoring of fishing vessels activities.
Authors: Danhiez, Francois, Weiss, Jérôme, Rodriguez, Julien, Woillez, Mathieu
Language: English
Subfield: Aquatic Science
Domain: Life Sciences
SDG: Life below water
Open Access status: green
OpenAlex ID: W7080053949
