Hybrid Black Widow Combined Seagull Optimization with Deep Learning for Efficient Road Classification in UAV-Aided Intelligent Transportation Systems

Unmanned Aerial Vehicles (UAVs) are pivotal in Intelligent Transportation Systems (ITS) for smart cities, enabling interconnected and autonomous vehicle networks.UAVs enhance ground vehicles by establishing efficient wireless connections and aiding in real-time road monitoring.Artificial Intelligence (AI) and Machine Learning (ML) optimize UAV operations, including dynamic control, path planning, and environmental perception.Deep Learning (DL), particularly Convolutional Neural Networks (CNNs), excels in detection tasks, making it ideal for UAV-based road classification.This study proposes a Hybrid Black Widow Combined Seagull Optimization with Deep Learning-based Classification (HBWCSO-DLC) algorithm to improve road classification accuracy.The HBWCSO-DLC integrates Black Widow Optimization (BWO) and Seagull Optimization (SOA) to enhance feature selection and CNN performance.Evaluated on a road image dataset, HBWCSO-DLC achieves 99.48% accuracy, 98.69% sensitivity, 99.67% specificity, and a 98.69% F1-score, outperforming existing methods like MODAE-RCM, Adam, and SGD.The hybrid optimization ensures robust convergence, while the CNN architecture adapts to complex road textures captured by UAVs.The results demonstrate HBWCSO-DLC's superiority in ITS applications, including autonomous driving and safety systems.This work provides a scalable solution for UAV-assisted road classification, combining bio-inspired optimization with deep learning for real-time, high-precision outcomes.