Answers are grounded in this paper and cite their sources.
Pavement crack analysis is a representative benchmark problem for intelligent visual inspection, requiring detection of thin, low-contrast, topologically irregular structures under severe class imbalance. Existing surveys remain fragmented across levels of abstraction, typically addressing algorithmic performance, data acquisition, or management systems in isolation. This survey provides a system-oriented synthesis based on a unified taxonomy with three functional layers: (i) data acquisition and sensing; (ii) crack detection and analysis algorithms, from classical vision to advanced Deep Learning, including foundation models, Large Language Models (LLMs), and Vision-Language Models (VLMs); and (iii) downstream applications, comprising predictive maintenance and Digital Twins. Public datasets and evaluation practices are reviewed, examining how task formulation, annotation granularity, sensing modalities, and acquisition conditions affect evaluation validity and generalization. Computational and deployment considerations (efficiency, interpretability, scalability, and robustness) are then discussed to guide the development of pavement crack analysis solutions that are both scientifically rigorous and practically deployable.
No takes yet. Share an insight, caveat, or question.
Castillo-Secilla et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: