Detecting Underwater Macroplastic Pollution: A Review on AI-Based Computer Vision Methods and their Relevance to Sri Lanka’s Marine Ecosystem

Abstract

The escalating crisis of marine plastic pollution, particularly macroplastic debris, poses severe threats to marine biodiversity, coastal ecosystems, and human livelihoods. Sri Lanka, ranked among the top global contributors to marine plastic waste, suffers from a lack of systematic and scalable monitoring mechanism, especially for underwater macroplastic detection. Traditional survey methods are often manual, labour-intensive, and limited in coverage. Recent advancements in artificial intelligence, specifically computer vision, present promising alternatives for automating detection and analysis of marine debris. This review paper critically examines the current landscape of AI-based approaches for underwater macroplastic detection, highlighting their methodologies, datasets, performance metrics, and limitations. While global efforts have largely emphasized surface litter and microplastic detection, underwater macroplastic monitoring remains underexplored. Moreover, Sri Lanka lacks context-specific studies and affordable, software-based AI tools that can leverage existing underwater imagery to detect macroplastic waste. By synthesizing global literature and identifying region-specific research gaps, this paper advocates for the development of lightweight, AI-powered systems tailored to the Sri Lankan coastline. Such innovations could enable continuous monitoring, inform policy decisions, and strengthen marine conservation efforts in data-scarce and resource-constrained settings.