Spatial analysis and complexity evaluation for predicting rebar installation duration

Poor predictability of construction project duration and stagnant labor productivity highlight the need for scalable, data-driven monitoring solutions. This paper investigates how the relationship between activity complexity and duration can be quantified and leveraged. A grid-based spatial analysis is introduced to address this gap, combining BIM-derived complexity metrics for individual components (e.g., slabs) with CV-based as-performed activity classification, demonstrated in the context of rebar installation. The proposed activity classification model, ViTPoseActivity , achieves 97% accuracy in detecting on-site tasks using single-frame posture features. Correlation analysis on real-world construction data covering over 1,900 labor hours confirms a measurable positive relationship between activity complexity and duration. By combining as-designed and as-performed data in a spatial context, this paper provides a foundation for activity duration prediction, supporting proactive planning and future research in data-driven site management. • Grid-based spatial analysis quantifies activity complexity-duration relationships in construction. • CV-based activity classification architecture (ViTPoseActivity) achieves 97% detection accuracy. • BIM-derived complexity metrics enable localized correlation analysis for proactive planning. • Analysis of 1,900+ labor hours in rebar installation confirms measurable activity complexity-duration correlation.