Abstract
Global fisheries face an urgent crisis of biodiversity loss due to the unintentional capture of non-target species, known as by-catch. Traditional Electronic Monitoring (EM) systems rely on manual review of video footage, often resulting in data backlogs and delayed management responses. This research explores the integration of Edge AI—high-performance, localized artificial intelligence—onboard industrial fishing vessels to automate species identification and quantification in real time. By deploying state-of-the-art computer vision models, such as YOLOv8 and ShuffleNetV2, on ruggedized hardware like the NVIDIA Jetson platform, vessels can now detect by-catch at the point of capture (Lee Son et al., 2023). This paper analyzes recent industrial trials where Edge AI reduced footage review times from months to minutes, achieving catch-count accuracy with miss rates as low as 6% (Responsible Seafood Advocate, 2026). We discuss the technical architecture required for maritime environments, including preprocessing filters and satellite-linked reporting, and demonstrate how this "human-in-the-loop" technology enhances sustainable fisheries management and transparency.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Copyright (c) 2026 Scholarly Journal of Science and Technology Research and Development