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
Autonomous Rail Surface Defect Detection
-
2024-09-30
-
[PDF-1.08 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:June 1, 2023 – August 31, 2024
-
Corporate Publisher:
-
Abstract:The Autonomous Rail Surface Defect Detection project aims to enhance railway safety through the use of unmanned aerial vehicles (UAVs) for detecting rail surface defects. Using the RSD_UAV dataset, the project developed an optimized DeepLabv3-plus model with a ResNet-18 backbone and CBAM, achieving a mean Intersection over Union (mIOU) of 84.97% and a mean accuracy of 92.60%. The dataset, containing 13,053 images of various rail defects, was collected in Columbia, SC, and rigorously processed for model training. The system's performance was tested across different UAV flight patterns, revealing consistent defect detection but with reduced accuracy as the UAV's altitude increased. Optimal detection occurred between 3 to 9 ft above the rail, with accuracy decreasing at higher altitudes or lateral distances from the rails. This research underscores the potential of UAVs and deep learning models in advancing railway inspection and safety.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
