United States Department of Transportation
ROSA P serves as an archival repository of USDOT-published products including scientific findings, journal articles, guidelines, recommendations, or other information authored or co-authored by USDOT or funded partners.
As a repository, ROSA P retains documents in their original published format to ensure public access to scientific information.
i
Deep Learning-Based Computer Vision Framework for Predicting Retroreflectivity of Road Signs
-
2024-09-01
-
[PDF-1.12 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report July 2024
-
Corporate Publisher:
-
Abstract:The retro-reflectivity of traffic signs is essential for road safety, particularly in low-light and adverse weather conditions, by ensuring signs remain visible and legible to road users. However, retro-reflective quality degrades over time due to environmental exposure, reducing sign effectiveness and increasing risks of road accidents and potential liabilities for road agencies. Accurate estimation of traffic signs’ lifespan is thus critical for efficient maintenance and resource allocation. While prior studies have analyzed factors affecting retro-reflectivity, such as sign age, color, orientation, and weather, few have utilized deep learning models with expanded input variables. In this study, we address these gaps by examining additional environmental factors, air temperature, relative humidity, and sign age through a deep learning model incorporating image detection. The results of the XGBoost classification model achieved an overall accuracy of 0.97, correctly predicting 97% of instances, while our transfer learning model, EfficientNet-B0 results yielded an R² value of 0.79, indicating that 79% of retro-reflectivity variance was explained, highlighting the model’s capability to identify signs that fall below retro-reflectivity standards. The model’s performance highlights the potential for automating the detection of road signs below required retro-reflectivity standards, reducing maintenance costs, and enhancing road safety through predictive monitoring.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
