Project Summary
As the nodes of transportation networks, where modal and directional traffic flows convene, intersections are inherently complex and critically important. Occupying only a small share of urban streets, intersections account for a disproportionate number of traffic fatalities and represent one of the most inconsistently designed elements in U.S. streetscapes. Local governments face growing pressure to improve safety and accessibility, yet lack scalable tools to systematically assess intersection performance, compare designs across jurisdictions, or identify what works. We introduce CrossWise, a context-aware platform for intersection retrieval and comparison. By leveraging computer vision, spatial analysis, and graph-based learning, CrossWise embeds intersections into a shared feature space that captures both physical form and surrounding context. This enables cross-jurisdictional analysis of form-function relationships at scale. CrossWise empowers planners to identify designs associated with safer, more equitable outcomes, detect infrastructure disparities, and support initiatives such as Vision Zero, Safe Routes to School, and network-wide investment prioritization.
