Artificial Intelligence–Driven Computer Vision in Construction Industry: A Systematic Review of Applications and Trends

Abstract

The rapid evolution of computer vision (CV) and artificial intelligence (AI) is reshaping civil engineering by enabling automated perception, monitoring, and decision-making across the project lifecycle. However, the research landscape remains fragmented across applications, technologies, and engineering domains, limiting a consolidated understanding of its maturity and future trajectory. This study addresses this gap through a systematic review of 721 journal articles indexed in Scopus, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework. Bibliometric, thematic, and qualitative analyses were integrated to characterize the evolution, application domains, technological trajectories, and persistent research gaps in CV-based civil engineering research. The findings reveal a marked transition from exploratory applications before 2015 to accelerated growth after 2018, coinciding with advances in deep learning and computational capabilities. Research is strongly concentrated in construction management and safety, particularly equipment recognition, worker monitoring, personal protective equipment detection, hazard identification, defect detection, and progress monitoring. In contrast, geotechnical, water resources, and environmental applications remain comparatively underdeveloped. Emerging convergence among CV, deep learning, BIM, UAVs, and digital twins indicates a shift from isolated perception tasks toward integrated intelligent construction systems. Nevertheless, limited fieldscale validation, inadequate cross-site generalization, dataset heterogeneity, and insufficient benchmarking constrain practical deployment. The review synthesizes these limitations into a research agenda emphasizing standardized datasets, transferable models, multimodal integration, and field-validated intelligent systems.