A computer vision-based approach to monitor changes in ecosystems associated with marine renewable energy projects

Abstract

Abstract To assess the potential ecological effects of marine renewable energy (MRE) devices, it is essential to establish a baseline of ecosystem conditions prior to disturbance and to monitor throughout the installation, operation, maintenance, and decommissioning phases. The design of mitigation and compensation measures depends on the difference between baseline conditions and the impacted ecosystem. In this study, we develop and evaluate a semi-automated methodological proof-of-concept using computer vision to assess environmental changes around MRE devices. Using fish monitoring as a case study, we reduced image processing time by eliminating empty frames and detecting fish in the remaining frames. Species identification was carried out by experts using a public database for comparative analysis. We also conducted an identification exercise for a single fish species and tested a monocular depth-estimation method (3D reconstruction from 2D images) to measure distances between organisms and devices. Finally, we propose a potential methodological framework for integrating computer vision across the life cycle of MRE devices to monitor both natural and anthropogenic processes, including space colonization, trophic interactions, pollution, and the effects of underwater structures. The merits and drawbacks of the approach are discussed.