Winner of Best in Show Fall 2020 Senior Design Showcase.
Cemetery Drone Image Analysis is a computer vision system that automatically locates and maps individual headstones in aerial photographs of national cemeteries. Sponsored by Dr. Amy Giroux, the project grew out of a real bottleneck in her work: she flies a drone over national cemeteries to capture detailed overhead imagery, then stitches the photos into a single large image of the grounds—but identifying each headstone meant manually drawing a polygon around every single one, an exhausting task for cemeteries spanning more than ten acres with hundreds of graves. This project automates that process using a machine learning model (a Mask R-CNN trained on Giroux’s drone imagery and supplemental Google Earth data) that recognizes each headstone, outlines it with a polygon, and uses the cemetery’s geographic data to calculate precise latitude and longitude coordinates for every marker. Those results feed into a SQLite database that catalogs each headstone by location, turning a slow manual workflow into a fast, automated mapping tool that helps preserve and organize the records of those buried there.