Integrating Drones and Deep Learning for Automated Detection of Invasive Species

This study presents an innovative approach that combines drone-based image processing, deep learning and GPSassisted mapping systems for the detection and management of invasive plant species. In the research, high-resolution images collected by drones are analyzed using the YOLO algorithm and each invasive plant species is mapped onto geographic maps with centimeter-accurate RTK-assisted GPS data. The visualization clearly distinguished the spatial distribution of each species with color codes, allowing easy identification of areas where invasive species are concentrated. The results revealed areas where invasive species such as vetch, kanyash and goosefoot are densely populated and provided important data to understand the distribution dynamics of these species. Harmless species were observed to act as a natural barrier against invasive species, indicating that agricultural activities can be an effective method for invasive plant control. The algorithm performed well with 72.2% accuracy, 74.2% precision and 78.4% sensitivity. This method not only enabled effective detection and management of invasive species, but also demonstrated the possibilities of technology for sustainable management of agricultural landscapes. The study makes an innovative contribution to the field of agricultural and environmental management by providing a fast, reliable and cost-effective solution.