2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference


Title
Big Visual Data Analysis for Building Information Modeling

Return to Session
BIM

Author(s)
Dr. Cheng Zhang, Xi'an Jiaotong-Liverpool University (Presenter)
Fangyu Guo
Dr. Lei Fan, Xi'an Jiaotong-Liverpool University
Mr. Yousif Ali
Abstract

The Big Data concept is now receiving remarkable attention for tackling complex engineering problems. Among different engineering fields, Big Data analytics is notably impacting the Civil Engineering domain. However, despite the significance of the Big Data technologies to process large-scale data, current Civil Engineering information systems are still lacking in successful implementation of them. Big Data include all kinds of data that can be collected by different means, such as sensors, cameras, laser scanners, etc. Among those big data, visual data have better accessibility and have more impressions on human being’s daily life. The ever increasing volume of visual data due to recent advances in smart devices and camera-equipped platform provides an unprecedented opportunity to visually capture actual status of the physical environment at a fraction of cost compared to other alternative methods. This provides an unprecedented opportunity to understand the construction processes, but also requires advanced management skills and data-processing ability.

This paper investigated methods to collect on-site data using different technologies, and how those data can be fused into an information modeling platform, including geometric and sematic information. During the data acquisition stage, key questions will be investigated and answered. For example, What to collect? How to collect? In which form? For what purpose? How to use? A sequential and automatic framework is developed to analyze those data, which includes Image Registration, Image Feature Extraction, Object Shape Detection, and Object Updating/Tracking (static/moving object). Meanwhile, semantic meanings will be integrated into the information model, where the data collected as attributes will be processed and associated with corresponding components in the geometric models. Semantic enrichment of building models adds meaningful domain-specific or application-specific information to a digital building model. Algorithms are developed to identify different materials and components based on data collected. A case study is applied on a construction site by collecting visual data and integrating the processed information into a BIM model. Preliminary results show that more efforts should be contributed in building a roadmap based on big visual data analysis, where raw data need to be filtered, extracted and summarized into domain-related information and build information models for the entire built environment.