Leveraging DesignSafe for AI-Driven Hazard Research: A Debris Segmentation Case Study

About the Webinar

There is a growing interest in applying and fine-tuning foundation models in the broader natural hazards engineering community, and the DesignSafe Cyberinfrastructure offers significant potential to leverage these advanced AI tools. This webinar will explore how DesignSafe’s integrated ecosystem—spanning data publication, computational resources, and shareable tools—can empower researchers to develop and apply these powerful models.

We will use a recent research project on post-disaster debris segmentation as a practical case study. This project highlights the full research lifecycle on DesignSafe: starting with the curation and publication of a new, benchmark dataset of post-hurricane aerial imagery, which is now publicly available on DesignSafe (PRJ-6029). We will then demonstrate how this dataset was used to fine-tune a large-scale vision model (CLIP) to create a robust, generalizable tool for identifying debris. The webinar will focus on the practical application and deployment of these tools using DesignSafe, featuring a comprehensive computational notebook. We will also discuss opportunities and potential workflows for leveraging TACC’s powerful computing resources for large-scale model training and inference.

Join us on Wednesday, December 3rd, 2025, to discover how you can use DesignSafe to advance your own AI-driven natural hazards research.

Presenters

Jamie Ellen Padgett
Stanley C. Moore Professor and Chair
Department of Civil and Environmental Engineering, Rice University, Houston, TX, USA
Ken Kennedy Institute, Rice University, Houston, TX, USA

Kooshan Amini
Ph.D. Student
Department of Civil and Environmental Engineering, Rice University, Houston, TX, USA