Large-scale remote monitoring of riverine litter using drones and bridge-mounted cameras.

The INSPIRE Horizon Europe project aims to combat plastic pollution in rivers by deploying 20 scalable technologies designed to prevent and remove litter. These innovative solutions will be demonstrated across six rivers in Europe. A key component of the project involves monitoring plastic pollution in the rivers and on their banks to establish a baseline and assess the effectiveness of the implemented technologies.

One of the monitoring solutions introduced is a camera- and drone-based system for tracking plastic flux in the rivers and measuring macroplastic densities on riverbanks. The fixed camera system utilizes a series of Commercial Off-The-Shelf (COTS) surveillance cameras, equipped with protective housing and real-time data links. These autonomous cameras covering the full width of the river, continuously upload data to the cloud, providing uninterrupted monitoring. In addition, the drone data in collected along the riverbanks using high-resolution RGB - camera.

Robust machine learning models, including convolutional neural networks (CNNs) and advanced pre-trained foundation models like Segment Anything, are employed for litter detection and characterization. Furthermore, innovative approaches are being explored to transform camera-based plastic detections into plastic flux measurements. These methods leverage feature detection techniques such as Faster R-CNN, pretrained on ImageNet and further fine-tuned for this purpose, combined with feature matching algorithms like ResNet and auxiliary data integration.

The presentation will cover preliminary findings from fixed camera systems installed on Belgium’s Temse Bridge, as well as drone data collected along the Scheldt riverbank (BE) and in the Londenhaven area of the Port of Rotterdam (NL).