Todd Matuszewicz
AI Large Visual Model for Identifying Historic Properties:
Pioneering Surveying and Building Identification Efforts in Denver
Creating a Modern Digital Sanborn Map
Todd Matuszewicz is a fifth-generation Denverite living six blocks from where he was born. After receiving a bachelor’s degree in chemistry with a minor in history from Metropolitan State University of Denver, he worked as a neon tube bender for more than a decade before transitioning to a 15-year career as an educator at the Denver Waldorf School. Returning to the neon trade in 2020, Todd focused on the restoration of and preservation of neon signs. This transition back toward the trades led him to enroll at the University of Colorado Denver’s Dana Crawford Historic Preservation program where he is a Master’s candidate. While pursuing his master’s degree, Todd works as a historic building researcher for Historic Denver and restores neon at Morry’s Neon. Todd’s advocacy for vintage neon sees him lecturing at conferences, presenting on webinars, and hosting tours which conclude with participants trying their hands at glass bending. In 2024, Todd cofounded Save the Signs, Colorado, an organization devoted to preserving signs in situ.
Todd’s fellowship capstone project is titled, AI Large Visual Model for Identifying Historic Properties – Pioneering Surveying and Building Identification Efforts in Denver (working title: Creating a Modern Digital Sanborn Map). The Creating a Modern Digital Sanborn Map project aims to expedite the ability of cities to survey and document their cultural resources utilizing emerging technology.
Historic Denver’s Discover Denver program is a trail-blazing effort to document every building in Denver, approximately 160,000 in total. The process relies on volunteers to photograph and manually enter pertinent architectural and historical features. The data is quality-checked by professionals who identify potential properties for further research. The team has reviewed nearly 40,000 cases in 10 years. Matuszewicz will utilize this invaluable data to create a machine-assisted large visual model (LVM), an Artificial Intelligence agent, to perform the initial analysis. The goal of the Creating a Modern Digital Sanborn Map project is to introduce a working agent to the Discover Denver process and reduce the project’s completion date from thirty years to eight years. Supported by a multidisciplinary team of professionals, Todd will develop an open-source AI model that can be easily trained to recognize not only other regionally distinct architectural styles, but to be trainable to survey non-building resources: i.e. signs, roadside colossi, and culturally modified trees.
Todd plans on demonstrating the final iteration at conferences and on webinars, will create a how-to booklet and instructional YouTube channel, offer guest lectures in university preservation programs nationwide, and showcase the tool for Registered Neighborhood Organizations pursuing Landmark District status. Team members have also discussed the possibility of a phone app allowing individuals to survey their property.
Knowledge is power. The world will be guided, not by AI, but by those who know how to utilize it. In collaboration with the Harrison Goodall Fellowship team to facilitate surveying innovation, Todd is committed to the fundamental values of preservation maintaining a place in shaping our shared future.
