A Robust Vision Based Lane Scenario Detection and Classification Using Machine Learning for SelfDriving Vehicles

In the five levels of autonomous vehicles, fully autonomous driving is an essential requirement where lane detection is one of the critical challenges. With zero human interaction for lane detection, a high safety margin must be considered, requiring complex decision-making algorithms. In that case, Machine learning algorithms are the highly trusted option. In this paper, a simplified camera-based lane detection and scene classification method is proposed depending on different road conditions so that autonomous vehicles can operate in various challenging scenes with a single camera and accurately recognize lanes under any conditions. Our proposed instance segmentation-based lane detection and classification algorithm can identify different road conditions by analyzing different scenes in complex situations. The training process is enriched with in-field data set which delivers a significant accuracy.