A computer vision and reasoning based video interpretation system for rapid productivity analysis of construction operations

On construction jobsites, considerable manual data collection and analyses efforts are often required to conduct productivity studies. Intelligent computer vision systems that can autonomously observe the behavior of construction objects and infer their working status have the potential to enable automated productivity assessment and therefore reduce some of the shortcomings of the manual methods. As an important step towards developing field deployable intelligent vision systems, this research addresses the problem of automated productivity data collection through the development of an intelligent construction video interpretation methodology and the implementation of an intelligent video computing system based on the developed methodology. The proposed methodology was validated through using the developed prototype system to analyze five construction video sequences that record various types of construction operations. The developed system was able to interpret these videos into productivity information with an accuracy that was close to manual analysis, without the limitations of onsite human observation.