Action reliability assessment framework for automated construction labor measurements: Case study on plastering operations

Automated methods for collecting data on manual labor significantly enhance efficiency and reduce costs in the assessment of labor productivity. A critical aspect of productivity analysis involves identifying effective construction actions from a wide range of human activity data. Most existing approaches have concentrated on the biological characteristics of human actions to classify effective construction activities; these methods rarely evaluated whether these actions lead to meaningful labor outcomes and the reliability of automated measurement techniques remains a challenge, necessitating further refinement to ensure accurate and consistent results. This study presents an analytical framework designed to enhance the reliability of action recognition models by systematically pairing each effective construction action with its corresponding labor output. The innovation lies in the development of a structured action evaluation process that utilizes the spatial relationship between workers' movements and work objects. The framework is structured around three key components: worker position detection, action evaluation, and output validation. Central to this approach is the use of the construction object as a spatial baseline, which serves to quantify the worker's operational space and measure effective labor. Within this defined operational space, preliminary action identification is performed by analyzing the trajectories of key joints. Subsequently, the movement trajectories of these key joints are examined along the spatial baseline to estimate labor output, thereby establishing a spatiotemporal correspondence between each action and its resulting labor output distribution. By integrating these processes, the study introduces a novel method for assessing the rationality and effectiveness of construction actions. To validate the proposed approach, a case study focusing on the estimation of direct work in plastering operations was conducted. The results of this case study demonstrate that the proposed approach reliably verifies the accuracy and effectiveness of construction operations.