Vehicle Detection and Classification in Aerial Images with YOLOv8 and CNN-BiLSTM

The detection and classification of vehicles in complex traffic monitoring systems pose significant challenges that demand innovative solutions. Traditional methods often struggle with high computational requirements and the need to integrate diverse data sources. This research presents a strong framework particularly built for vehicle classification in aerial picture sequences. Our method comprises georeferencing, preprocessing, segmentation, and detection using the YOLOv8 algorithm, augmented by feature extraction utilizing BRIEF and FAST approaches. For the classification, we utilize a hybrid model that blends Convolutional Neural Networks (CNN) with Bidirectional Long Short-Term Memory (BiLSTM) networks, providing extensive analysis and assessment. By integrating these advanced methodologies, our framework significantly improves vehicle detection and classification, thereby enhancing traffic management and surveillance systems. Experimental results on the Vehicle Aerial Imagery from a Drone (VAID) dataset demonstrate that our approach outperforms existing methods, achieving an impressive accuracy of 0.93%. This demonstrates the efficiency of our method in recognizing and classifying cars in aerial imagery, representing a major advancement over past techniques.