The intelligent detection of pavement distress using deep learning methods has consistently been a hotspot in pavement maintenance. This paper aims to offer new insights to promote research and application in this field through bibliometric analysis. Utilizing publications from the Web of Science Core Collection spanning from 2016 to 2024 as the database, this paper conducts a systematic analysis of statistical data concerning the annual publication numbers, countries/regions, institutions, authors, hot papers, disciplines, and journals. Based on deep learning models, datasets, and the state of practice, this analysis explores the hotspots and fronts of this field. It identifies gaps, challenges, and future research directions, including the exploration and optimization of models, the quality and variability of datasets, the evolution of data acquisition methods, the impact of the state of practice, the prospects of unmanned detection technologies, the integration of multi-source heterogeneous data, and the potential of digital twin technologies.
