Title:
An automatic survey method for determining live loads based on multi-source Internet data and computer vision
Abstract:
The measured data of live loads form the basis for the structural design and analysis. This study focuses on the amplitude measurement of the sustained load, which is a basic type of the live load. Traditional survey methods are characterized by manual and on-site operation and lead to a series of problems including the high cost, low efficiency and occupant resistance. Taking full advantages of the unlimited Internet resources and computer vision technology, a new survey method is proposed to realize an automatic and online investigation into the load amplitude in residential buildings. The object weights and room areas are directly acquired from the commodity information on e-commerce websites and the residence attributes on real estate websites, respectively. An object detection model based on YOLOv4 algorithm is developed to identify the object quantities from the open-access room photos on real estate websites. The detection model is trained by using a dataset of 8000 room photos. The performance of the detection model is evaluated on a test dataset including 1000 room photos. Finally, object quantities in 343 rooms are obtained by manual counting and model recognition, respectively. The difference between the manual and automatic survey results is no more than 15%, which verifies the feasibility and accuracy of the proposed method.
Bio:
Chi Xu is a doctoral candidate at Tongji University, under the guidance of Professor Jie Li. He obtained his Bachelor’s degree from Tongji University in 2019 and was recommended for direct admission to the doctoral program. His research focuses on live load modeling, the application of Big Data in civil engineering, computer vision based on Deep Learning, and load combination methods. He has presented his research findings at several international academic conferences and has contributed to the publication of 11 academic papers, including those currently under review.
