Marked crosswalks in US transit-oriented station areas, 2007–2020: A computer vision approach using street view imagery

This is the supplementary dataset for the article “Marked crosswalks in US transit-oriented station areas, 2007–2020: A computer vision approach using street view imagery” (https://journals.sagepub.com/doi/10.1177/23998083221112157) published on Environment and Planning B: Urban Analytics and City Science. The dataset contains comprehensive information of the presence of marked crosswalks (i.e., parallel-line and high-visibility) at each street intersection within a 250-meter buffer of all US TOD stations for each consecutive year between 2007 and 2020. We describe each field of the dataset below. Most of the column station variable names are taken directly from the National TOD Database from the Center for Transit-Oriented Development. 

intersecton_id: numeric, unique identifier of each intersection.

tod_id: numeric, unique identifier of the associated TOD station. We retain all records of intersections that are associated with multiple TOD stations. 

year: numeric, one year between 2007 and 2020. 

plain: numeric, the number of presence of parallel-line crosswalks (bounding boxes) detected among the four 90-degree images at each intersection in that particular year. Zero for the years where GSV are not available. 

zebra: numeric, the number of presence of high-visibility crosswalks (bounding boxes) detected among the four 90-degree images at each intersection in that particular year. Zero for the years where GSV are not available. 

impute: binary, 1 if imputed; 0 otherwise. 

lon_intersection: longitude of each intersection obtained from OSM. 

lat_intersection: latitude of each intersection obtained from OSM. 

distance_intersection_to_tod: distance of each intersection from TOD station by meters. 

Agency: operating transit agency of the associated TOD station. 

Lines (s): transit lines of the associated TOD station. 

Station Name: station name of the associated TOD station. 

Year Opened: the year when the associated station opened. 

lat_tod: latitude of the associated TOD station obtained from the National TOD Database. 

lon_tod: longitude of the associated TOD station obtained from the Nation TOD Database. 

Both: numeric, the total number of marked (parallel-line + high-visibility) crosswalks detected among the four 90-degree images at each intersection in that particular year. 

plain_b: binary, 1 if presence of parallel-line crosswalk; 0 if otherwise. 

zebra_b: binary, 1 if presence of high-visibility crosswalk; 0 if otherwise. 

both_b: binary, 1 if presence of marked crosswalk; 0 if otherwise.