A survey on traffic violations prediction with deep learning

Abstract

Despite growing interest in traffic violation prediction, there is a lack of comprehensive survey research on this topic. A systematic survey is essential to understand the current state-of-the-art methodologies and to identify promising directions for future work. This paper surveys research on traffic violation prediction from the past five years, with a particular focus on machine learning and deep learning approaches. It provides an in-depth analysis of model architectures, data characteristics, and the types of traffic violations addressed in existing studies. In addition, this survey highlights current challenges, underrepresented violation types, and methodological best practices. Finally, it explores possible opportunities for future research, including possible integration with other domains such as gamified intervention.