Singano, Afika2026-03-172026-03-172026-012581-5598http://hdl.handle.net/10204/14770Satellite imagery analysis has become increasingly important for various applications, including urban planning, infrastructure development, transport asset management and disaster response. One critical task in satellite image analysis involves the detection and classification of roads. In this study, an approach utilizing the YOLO (You Only Look Once) Convolutional Neural Network (CNN) object detection model is proposed to distinguish between paved and unpaved roads from satellite imagery. Leveraging a custom dataset curated for this purpose, a YOLO object detection model was trained on Google Colab Pro's infrastructure, achieving promising results. Our methodology offers a robust and efficient solution for road type detection, with potential applications in urban development, transport planning, and environmental monitoring.FulltextenMachine learningPaved roadsRemote sensingRoad type classificationSatellite imagery unpaved roadsYOLOv8Articlen/a