Indoor construction productivity assessment using computer vision and mask region-based convolutional neural networks

Purpose – Current manual inspection and monitoring practices of construction productivity (CP) for indoor finishing activities are time-consuming and dependent on the experience of inspectors. As the construction industry undergoes a transformative phase amidst the emergence of the fourth industrial revolution, there is a growing exploration of technological solutions aimed at enhancing the monitoring process for CP. The purpose of this study is to develop computer vision model for CP assessment of the final phase of bricks laying, plastering, painting and tiling activities. Design/methodology/approach – The methodology involves developing a Mask Region-Based Convolutional Neural Network (Mask R-CNN) architecture, leveraging its instance segmentation capabilities to classify indoor construction activities. Based on the classification outcomes, a model is designed to calculate the area of accomplished activities based on the segmented regions in the images for the purpose of CP assessment. Findings – The CP assessment model achieves an accuracy of about 97% for bricks laying, 94% for plastering, 96% for tiling and 93% for painting. Regular assessing using the developed model serves as a proactive approach for identifying potential issues and tracking adherence to project timelines. Originality/value – The contribution of this study lies in its development of CP assessment model for indoor finishing activities. The presented model enhances efficiency in project management and contributes valuable insights to the construction industry and construction management literature. The integration of Mask R-CNN for CP assessment represents a significant development of advanced technologies in revolutionizing traditional construction monitoring practices.