Artificial Intelligence Framework for Crosswalk Detection Across Massachusetts

Details

• Creators:
  Apostolov, Atanas ; Xie, Yuanchang ; Ai, Chengbo ; Tainter, Francis ; Oke, Jimi
• Corporate Creators:
  University of Massachusetts at Amherst ; University of Massachusetts at Lowell
• Contributors:
  Polin, Bonnie
• Corporate Contributors:
  Massachusetts. Department of Transportation. Office of Transportation Planning ; United States. Department of Transportation. Federal Highway Administration
• Subject/TRT Terms:
  • Crosswalks
  • Artificial intelligence
  • Safety
  • Aerial photogrammetry
  • Intersections
  • Driveways
• Publication/ Report Number:
  20-024
• Resource Type:
  Tech Report
• Geographical Coverage:
  Massachusetts ; United States
• Edition:
  Final Report - February 2024 May 2023 - February 2024
• Corporate Publisher:
  Massachusetts. Dept. of Transportation. Office of Transportation Planning
• Abstract:
  The goal of this project was to develop an artificial intelligence (AI) framework to detect crosswalk locations across the state of Massachusetts, as well as their type classification (continental, parallel lines, or solid) and location category (intersection, midblock, or driveway). Aerial images downloaded from MassGIS were annotated, and these were used to train the AI model, which was then used to detect crosswalks across the entire state using images from both 2019 and 2021. About 88,000 crosswalks were detected in 2021. In terms of location category, 89% of these were intersection crosswalks, 8% were midblock, and 3% were at driveways. In terms of type, just under 65% were continental (zebra-style) crosswalks, about 35% were parallel lines, and less than 1% were solid/painted crosswalks. A python script was developed to perform the post-processing. The model and the post-processing script will be available to MassDOT to perform further analyses, and results will be expected to inform maintenance and safety initiatives.
• Format:
  PDF
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/75247/dot_75247_DS1.pdf
• File Type:
  [PDF - 2.84 MB]
• Collection(s):
  Artificial Intelligence Deployment Center of Excellence (AI COE) ; US Transportation Collection
• Main Document Checksum:
  urn:sha-512:f9bd42ad7e834f2a4e33199499f914b3190cb7eda438e747a9e0f1cf169ed4044da27c115502b1b93b33342fddbc60d62cb34bbf88523532eb12db95009d7ed0