Advancements in computer vision for safer overtaking: a review of deep learning methods

Road traffic accidents are a common global issue, causing injuries, fatalities, and substantial economic losses. Additionally, overtaking a vehicle is one of the leading causes of car collisions. Therefore, ensuring the safe overtaking of vehicles is a critical concern. With the advancement of Artificial Intelligence and its implementation in vehicles, many solutions have been proposed to tackle this problem. This review article explores image processing and deep learning techniques that enhance the safety of overtaking on roadways. It provides a comprehensive overview of methodologies and advancements in computer vision, mainly focusing on using deep neural networks to analyze and interpret real-time visual data to facilitate taking efficient and secure overtaking decisions in vehicular scenarios. It also examines traditional approaches to ensuring safe vehicle overtaking maneuvers and highlights their inherent limitations. Subsequently, the article delves into the important role of deep learning techniques in recognizing potential risks of vehicle overtaking, which helps make driving safer. Furthermore, the review discusses possible future directions in this field and identifies critical areas that require further research.