Using computer vision and street view images to assess bus stop amenities

Abstract

The assessment of bus stop amenities is important for providing fundamental data for public transit research, planning, and infrastructure enhancements. So far, public data on the amenities at bus stops have largely been unavailable. This study develops an automated, low-cost, and generalizable approach using Google Street View images and deep learning techniques to evaluate bus stop amenities. Leveraging the latest YOLOv8 model, transfer learning, and a dynamic prediction algorithm, our approach achieves efficient detection of shelters and benches with high accuracy and precision in major Florida cities. Results reveal highly heterogeneous spatial patterns for both shelters and benches within and across cities. Additionally, we conducted several tests to evaluate the transferability of the system to other urban contexts, which shows that highly accurate feature detection results can be achieved through model fine-tuning on a small sample of local data. In summary, the proposed system offers a scalable and efficient solution for large-scale real-time assessment of bus stop amenities, which can inform public transportation research and planning, especially for future transit infrastructure improvements.