zero-plastic: AI-assisted Sensing for Microplastic Assessment

Abstract Microplastics are widespread in aquatic environments and require continuous monitoring due to their high environmental and health risks. Accurate quantification remains challenging, as current methods rely on laboratory-based instruments that are expensive, labor-intensive, and unsuitable for large-scale or real-time assessments. This work presents the zero-plastic , a cost-effective, open-source AI-assisted imaging system for real-time microplastic monitoring, integrated with a planetary digital twin infrastructure. Built from accessible hardware and based on flow imaging microscopy, the system captures particles in the 3–12 µm range and processes images using an AI-based segmentation pipeline. Validation against scanning electron microscopy (SEM) shows good agreement for particles above 3 µm in size, confirming the system’s suitability for field-based monitoring. The device processes 0.3 mL per sample acquisition run and supports cloud-based data sharing. While particles below 3 µm are underdetected due to optical limits, the sensor performs reliably in its intended range. The platform enables continuous sampling and image analysis at low cost, with deployments underway in Brazil, South Africa, Ireland, and Scotland through the European Commission–funded ASTRAL project. This work contributes a reproducible, scalable tool for microplastic sensing in support of distributed environmental monitoring.