River Waste Detection Methods for Urban Canals in Southeast Asia—A Review of Present Techniques and Future Perspectives

Abstract

Floating plastic debris in urban canals is a management-relevant precursor to downstream microplastic pollution. This structured narrative review synthesizes conventional field surveys, camera-, UAV-, and satellite-based image analysis and AI-assisted image-based monitoring. Emphasis is placed on Southeast Asian engineered canals, where monsoon-driven flow, tides, gates, turbidity, glare, occlusion, and organic debris challenge continuous observation. Conventional surveys provide verifiable composition data but limited temporal coverage. Camera systems increase observation frequency, while deep learning can automate detection and tracking; however, reported performance depends strongly on the dataset, site, target size, and validation design. Published studies show substantial losses under cross-site transfer and condition-specific gains from preprocessing rather than a universal accuracy threshold. The synthesis therefore develops a decision-oriented framework linking camera calibration, conditional preprocessing, site-separated validation, uncertainty reporting, and hydrological data to operational triggers for cleanup or interception. Current evidence supports monitoring and pilot decision support, while broader autonomous operation requires further field validation. Priorities include transparent evidence reporting, shared Southeast Asian datasets, standardized metrics and environmental descriptors, cross-site testing, and life-cycle evaluation of deployment cost and maintenance.