Geospatial Artificial Intelligence for Solid Waste Recognition from UAV Imagery

Illegal waste disposal has direct consequences on people’s quality of life in affected areas. Exposure to hazardous waste substances has been linked to increased rates of multiple forms of cancer. Traditional inspection methods require time and manpower. Advances in UAV technology, combined with AI-enabled Computer Vision, offer a promising solution to significantly reduce survey time and the required workforce. The main contributions of this work are the evaluation of two Object Detection models, YOLOv8 and Faster R-CNN, trained to identify 7 waste materials from UAV imagery and the presentation of a practical pipeline to implement the proposed model in environmental agency workflows. To the best of our knowledge, no existing dataset or model has been designed to detect such a diverse range of waste types from UAV images. Results suggest that Object Detection is highly effective for regularly shaped waste categories such as Textile, Pallets and Tires, with the best YOLOv8 model achieving AP scores of 80.22%, 69.24% and 62.47% respectively. However, for waste materials with irregular boundaries, such as Rubble or Mixed Items, detection remains challenging. These findings provide valuable insights for future research in AI-driven environmental crime detection.