December 10, 2018
Assessing building damage is a critical first step in recovery post hurricane. Typically, damage is assessed in person – a time-consuming and laborious process. ISE PHD student Daniel Quoc Dung Cao proposed an alternative: apply an automated process to identify damaged buildings from satellite images. For his project, “Deep Learning Based Damage Detection on Post-hurricane Satellite Imagery," Daniel won an award at the Institute for Operations Research and the Management Sciences’ (INFORMS) 2019 Annual Meeting poster competition.
Working with Youngjun Choe, Assistant Professor of Industrial & Systems Engineering, Daniel used images of Hurricane Harvey to test his method. The researchers applied image classification techniques, in conjunction with geographic information system tools, to process and analyze the images. The method allowed for much quicker and more reliable identification of damaged areas. “We created the first dataset regarding damage detection that can be readily used by other researchers,” says Daniel. Daniel’s poster and presentation impressed the INFORMS poster competition judging committee. Highly competitive, the INFORMS competition accepts applicants worldwide from both industry and academia.
“It is a significant achievement for a second-year Ph.D. student to place in INFORMS’ poster competition,” says Professor Choe. “Daniel is a highly promising researcher, having already submitted two research articles to reputable journals in his first year while successfully completing heavy coursework. I look forward to his future achievements.”
Daniel plans to extend his research on hurricane recovery to identifying damage to transportation infrastructure. He hopes his research will aid in planning effective routes for food, medicine and other vital supplies to get to disaster survivors.