A systematic review of monocular depth estimation for autonomous driving: Methods and dataset benchmarking
Zhiwei Huang, Mohammed A. H. Ali, Y. Nukman, Haixia Xu, Shikai Zhang, Hui Chen, Mohammad A. AlKhedher · Results in Engineering · 2025
The utilization of monocular cameras for depth estimation in autonomous driving has emerged as a significant area of research, propelled by substantial advancements in artificial intelligence and the cost-effectiveness relative to other existing systems. This paper presents a comprehensive review of monocular depth estimation methods, which can be categorized into three primary approaches: Geometry-Based, Machine Learning, and Deep Learning. It encompasses numerous techniques proposed to enhance the accuracy and robustness of depth estimation from a single image. The principal contribution of this paper lies in its analysis of both traditional and deep learning approaches employed in monocular depth estimation, providing insights into their current state and historical context. Additionally, it offers a systematic categorization of the reviewed deep learning methodologies. The authors also discuss key innovations and limitations associated with each approach. Furthermore, common datasets and evaluation metrics utilized within this domain are examined, offering valuable insights into benchmarking practices that facilitate comparisons between various methods. Finally, this paper identifies open challenges and future research directions aimed at addressing issues such as scale ambiguity, dynamic objects, and generalization. This article provides an extensive overview of the field while fostering further advancements in monocular depth estimation.