AI-Based Road Lane Detection for Advanced Driver Assistance Systems: A Comparative Review of Traditional and Deep Learning Techniques

Road lane detection is an important part of Advanced Driver Assistance Systems (ADAS). It helps vehicles identify lane boundaries and maintain their position on the road. Traditional methods use imageprocessing techniques such as Canny edge detection, Region of Interest (ROI), and the Hough Transform. These methods are simple and fast but may have difficulty with faded lane markings, shadows, curved roads, and changing weather conditions. Recent AI-based methods use techniques such as CNN, YOLO-based models, image segmentation, and transfer learning to detect lane markings in complex environments, depending on the model architecture and training data. This paper reviews and compares traditional and AI-based lane detection techniques, with a focus on Indian road conditions. The comparison considers detection performance, processing speed, computational requirements, and robustness under different road conditions. Four selected research studies are reviewed to discuss the advantages and limitations of traditional and AI-based lane detection techniques.