Cost-Effective Traffic and Roadway Data Collection Using an Edge-Based Comprehensive Sensing System

Details

• Creators:
  Wang, Yinhai ; Shi, Yan ; Liu, Chenxi ; Chen, Kehua
• Corporate Creators:
  University of Washington. Washington State Transportation Center (TRAC) ; University of Washington. Department of Civil and Environmental Engineering
• Contributor:
  Jahangirnejad, Shervin
• Corporate Contributors:
  Washington State Department of Transportation
• Subject/TRT Terms:
  • Vehicle classification
  • Vehicle detectors
  • Deep learning
  • Expert systems
  • Data collection
  • Remote sensing
  • Cost effectiveness
  • Video cameras
  • multi-expert learning
• Publication/ Report Number:
  WA-RD 961.1
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States ; Washington
• Corporate Publisher:
  Washington State Department of Transportation
• Description:
  This project developed and evaluated a cost-effective, camera-based system for FHWA 13-bin vehicle classification using roadside video, edge computing, and deep learning. Traditional vehicle classification methods, such as inductive loops, pneumatic tubes, and manual counts, are often expensive to deploy and maintain and provide limited classification detail, particularly under challenging traffic, weather, and lighting conditions. To address these limitations, a portable roadside sensing platform was designed that integrates video cameras with edge computing hardware for on-site, near-real-time vehicle detection and classification. The system operates effectively in daytime, nighttime, and rainy conditions while minimizing data transfer requirements. At the core of the platform is an enhanced YOLOv11 deep learning model incorporating a mixture-of-experts (MoE) architecture. The model uses a shared backbone with multiple expert heads tailored to different environmental conditions, supported by a weather encoder and feature-wise linear modulation(FiLM) to adapt to variations in illumination, glare, and wet pavement. Model training and evaluation were conducted using a curated dataset of more than 20,000 vehicle instances annotated into the 13 FHWA vehicle classes and collected at WSDOT-relevant sites under diverse conditions. The training pipeline addressed class imbalance and visibility challenges through data augmentation, nighttime image enhancement, and class-weighted loss functions. The proposed system achieved an overall classification accuracy of up to 97.9 percent, with strong performance for heavy-duty vehicles and rare FHWA classes, even under adverse weather and lighting conditions.
• Format:
  PDF
• Funding:
  T1461-AI
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/93748/dot_93748_DS1.pdf
• File Type:
  [PDF - 4.97 MB]
• Collection(s):
  US Transportation Collection
• Main Document Checksum:
  urn:sha-512:2f0bf0921f22f88bc6c5075b1b0614d07756b356fe432891016521428daf0a7075f16f86f3f73da78f127d3556098b495fd37aceafea9bb210ebcfe735d34e57