Real-time Road Lane Boundary Monitoring System using Machine Learning

This paper presents a novel approach for lane detection using a Convolutional Neural Network with Line Detection (CNN-LD) methodology, aimed at enhancing the accuracy and efficiency of road lane recognition. The proposed model leverages advanced pre-processing techniques, including distortion correction, color space transformation, and noise reduction, to prepare input images for effective feature extraction. The methodology incorporates edge detection using Sobel filters and the Hough transform for precise lane identification. A comprehensive dataset of 4,000 annotated images captured under diverse lighting conditions—daytime, low light, and night-time—was utilized to train and evaluate the model. The CNN-LD framework demonstrated superior performance, achieving an accuracy of 98.92% and an F1-Score of 97.90%, significantly outperforming traditional methods. The integration of the Kanade–Lucas–Tomasi (KLT) tracker ensures robust lane tracking, even in challenging environments. Experimental results indicate that the proposed approach effectively addresses common issues in lane detection, such as variations in visibility and road conditions. This research contributes to the development of intelligent transportation systems, providing a reliable solution for autonomous driving applications. Future work will focus on improving model robustness against adverse weather conditions and integrating multimodal data for enhanced lane detection capabilities.