Visualizing Pedestrian Trajectories on a Map with MapLibre

This article describes a pipeline for transforming pixel-based pedestrian trajectories extracted from street video into geographic coordinates using homography, then visualizing them on an interactive map with MapLibre GL JS. The author uses Ground Control Points (GCPs) collected via a custom tool to compute the projective transformation from image space to WGS84. The project originated from a desire to measure pedestrian reactions during solo protests at Tokyo's Yamanote Line stations.