SignRecogGAN:Enhanced DCGAN for Traffic Sign Recognition Cyber-Physical Vehicular Systems

Autonomous Cyber-Physical Vehicular Systems (ACPS) combine physical components—such as mechanical systems and environmental interactions—with cyber elements, including computational and communication technologies, to create autonomous vehicles capable of interacting with their surroundings and making real-time decisions. A key focus of ACPS is ensuring road safety; one critical aspect is maintaining smooth traffic flow. Achieving this requires accurate detection of objects around the autonomous vehicle. This paper focuses on one of the essential applications of cyber-physical systems for autonomous vehicles: traffic sign detection. Traffic signs are a fundamental part of road infrastructure, and their proper recognition is crucial for safe and efficient autonomous driving. Here, we propose a deep convolution generative adversarial network rebalanced traffic sign recognition (SignRecogGAN) technique to address the problem above. This technique generates a new dataset through rebalancing, which is then used to compare the performance of models on both the real and the generated datasets. The generated dataset is then rigorously compared with the original real dataset using a machine learning model to evaluate performance metrics such as accuracy and precision. Our findings show that the model trained on the rebalanced dataset, using synthetic data generation, significantly improves classification performance for imbalanced datasets.