Intelligent Video Computing for Automated Construction Operation Productivity Analysis

My dissertation involved the development of video intelligence models for automated productivity analysis of construction activities. Gathering data for improving on-site operations is an essential and difficult task in construction. Among other things, video taping has long been used in construction to analyze construction operations. And recently, construction webcams have been increasingly used in construction projects for jobsite monitoring, providing live streams of visual information on ongoing construction activities. However, in the absence of an efficient video interpretation method, tedious manual reviewing is currently still required to extract productivity information from those videos. My research investigates intelligent computing methods for automated construction video analysis. The core of this research is to develop a generic computer-based video interpretation model that can interpret the live streams of visual information about ongoing construction activities and extract productivity information automatically at the level of human intelligence. The success of this research can greatly facilitate productivity data collection and real-time construction operation state intelligence. As a proof of concept, I have developed a model-based system that combines techniques from the areas of construction process modeling and simulation, construction productivity analysis, computer vision, and pattern recognition. The results demonstrated the feasibility and effectiveness of the proposed approach.