A Predictive Dashboard for Ocean Plastic Accumulation using Remote Sensing and Marine Meteorology

Marine ecosystems, biodiversity and coastal economies are all impacted by plastic pollution, which has grown to be a serious environmental problem. It is still challenging to monitor and forecast floating marine plastic on a large scale using conventional manual surveys and rule-based image analysis techniques. In order to identify and predict ocean plastic buildup, this research suggests an integrated intelligent monitoring framework that incorporates deep learning methods, marine meteorological parameters, and multispectral satellite imagery. To separate floating plastic debris from ocean features like waves, foam, and marine vegetation, a convolutional neural network based on U-Net performs pixel-level segmentation. A predictive drift modelling module uses wave characteristics, wind speed, and ocean current to estimate short-term plastic movement in order to facilitate proactive environmental monitoring. The real-time detection maps, hotspot visualisations, structural grid analysis, and temporal trend analytics are all provided by the system's interactive Streamlit-based dashboard. In comparison to traditional threshold-based segmentation and classical CNN approaches, experimental results show strong segmentation performance with high accuracy, precision, recall, IoU and dice scores. By facilitating automated monitoring, predictive risk assessment, and effective planning of plastic cleanup and environmental management strategies, the suggested framework offers marine conservation authorities a scalable decision-support system.