Work

Characterizing manufactured home parks in North Carolina: A computer vision based approach

Year: 2025

Type: article

Abstract: This study uses a computer vision-based approach to identify manufactured home parks (MHP) across North Carolina, addressing the critical lack of comprehensive public records for these vulnerable housing communities. Using high-resolution aerial imagery, a custom-trained YOLOv8 object detection model, building footprints, and parcel records, we identified 8460 MHP—more than three times the size of the largest current public dataset. The precision and recall of our model are over 93%, in detecting individual manufactured housing units. Results reveal that while large and medium-sized MHP are predominantly an urban phenomenon, small MHP (fewer than 50 units) constitute 95% of all identified parks, with over 67% of these small parks located in rural areas, and are concentrated in southeastern, central, and southwestern counties. This approach enabled comprehensive identification of MHP, supporting better-informed planning for these often overlooked but critical affordable housing options.

Source: Environment and Planning B Urban Analytics and City Science

DOI: 10.1177/23998083251395471

Authors: Kshitiz Khanal, Nikhil Kaza, Miyuki Hino, Antonia Sebastian

Institution: University of North Carolina at Chapel Hill

Language: English

Topic: Automated Road and Building Extraction

Subfield: Ocean Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Sustainable cities and communities

Open Access status: hybrid

Estimated APC paid: $3,650