Abstract
The present research explores the utilization of the Mask R-CNN convolutional network in the context of real-time road lane detection and segmentation, with a particular emphasis on its applicability to diverse environmental conditions. Through systematic experimentation and analysis, our study demonstrates the efficacy and adaptability of the proposed framework. The results, as illustrated in accompanying figures, highlight the model's capability to robustly identify and segment road lanes, thus contributing to the enhancement of autonomous vehicle navigation. Importantly, our system's proficiency extends to accommodating varied weather conditions, including scenarios with sunshine, rainfall, overcast skies, and others, as well as its adaptability to both day and night conditions. The real-time decision-making capacity of the system plays a pivotal role in optimizing the maneuvering of autonomous vehicles, offering potential benefits for road safety and traffic management. Moreover, we detail our dataset collection process and thorough pre-processing steps, ensuring the accuracy of model training and evaluation. This research contributes valuable insights into the practical application of deep learning, particularly the Mask R-CNN architecture, within the realm of autonomous driving, with implications for advancing road safety, traffic optimization, and the overall efficiency of autonomous vehicle operations, opening avenues for future refinements and expansions of this approach.