Enhanced Road Lane Marking Detection System: A CNN-Based Approach for Safe Driving

Accurate detection and interpretation of road lane markings (RLM) are essential for maintaining safe and efficient traffic flow. However, conventional computer vision methods encounter significant difficulties in handling occluded images. This research presents an innovative system named Road Lane Markings Detection (RLMD) that employs deep learning techniques to predict RLM from occluded images and provides their corresponding meanings. To address the limitations of existing approaches, a comprehensive dataset of RLM images, covering diverse occlusion scenarios, is created. Several deep learning baseline network architectures are used and trained on this dataset, enabling precise RLM prediction. Moreover, algorithms are devised to extract the meanings associated with various types of RLM, including broken lines, solid lines, and double solid lines in both yellow and white colors. Experimental evaluations demonstrate the system's effectiveness in real-world scenarios, showcasing its potential to enhance road safety and support driver assistance systems with 96.39% using convolutional neural networks (CNNs) layers. By overcoming the challenges posed by occlusions, the proposed system contributes significantly to the field of RLM prediction and its corresponding meanings using deep learning.