Deep Learning-Based Object Detection Techniques for Self-Driving Cars: an in-Depth Analysis

Self-driving cars are poised to transform transportation, but ensuring safe and reliable autonomous driving via robust object detection remains a critical challenge. This review comprehensively analyses object detection techniques within the self-driving car context. It highlights the significance of object detection focusing on the state-of-the-art deep learning (DL) based approaches. We investigate popular DL-based object detectors, assessing their strengths, weaknesses, and applicability to ensuring safety in self-driving scenarios. The paper explores the various network architectures, benchmark datasets, evaluation metrics and strategies to improve detection accuracy and efficiency. Moreover, we highlight unique challenges posed by self-driving environments and provide important research directions. Our findings underscore the ongoing, pivotal role of object detection as self-driving technology evolves. This paper equips researchers, practitioners, and policymakers with invaluable insights into shaping the future of transportation.