Damage Assessment through Aerial Imagery: A Deep Learning Approach in Disaster Recovery
This presentation discusses using deep learning approaches to improve disaster damage assessment through aerial imagery analysis. The researchers, from the University of Michigan working with the National Disaster Preparedness Training Center, explored three main technical approaches: image classification to categorize damage levels, object detection to identify specific features like blue tarps and debris, and change detection to compare pre- and post-disaster conditions.
While implementing these approaches requires significant computing power and high-resolution imagery, an argument can be made that the benefits of increased accuracy, efficiency, and consistency make it worthwhile. The presentation outlines both the technical requirements and challenges, such as access to quality imagery and computing resources, while providing recommendations for future implementation. The ultimate goal is to help emergency managers, assessors, and disaster response teams make faster, more accurate damage assessments to better serve communities affected by natural disasters.
Tab Troxler
Mr. Tab Troxler is an accomplished Homeland Security and Emergency Management professional with experience in both the public and private sector. As a former Emergency Services and Security Manager for the Dow Chemical Company, Mr. Troxler was responsible for the overall emergency management and response program for the site’s 2000-acre petrochemical manufacturing complex in Hahnville LA. In addition to his duties and responsibilities at the site, Mr. Troxler was a member of the Dow Gulf-Coast Hurricane team, as well as the Global Exercise and Plans Team. Mr. Troxler’s international assignments included responding as the Incident Commander for the Dow Soma Site in Japan following the earthquake and tsunami in 2011.
Previously, he served as the St. Charles Parish Director of Homeland Security and Emergency Preparedness where he led the parish’s response and recovery to Hurricane Katrina in 2005. As a former Senior Instructor with Louisiana State University’s, National Center for Biomedical Research and Training, Academy of Counter-terrorist Education, he has been on several development teams for national training programs in Computer Aided Management of Emergency Operations, Response to Weapons of Mass Destruction Cargo, Readiness: Training Identification Preparedness Training, and Food Management. He’s developed and facilitated several large-scale exercises internationally and domestically on behalf Louisiana State University. He currently is the elected parish Assessor in St. Charles Parish Louisiana. He received a Bachelor of Arts from Tulane University New Orleans, LA.