Transport agencies are increasingly adopting artificial intelligence and computer vision (CV) to augment traditional traffic infrastructure and improve operational outcomes. This presentation shares Auckland Transport's real‑world experience deploying camera‑based AI analytics in live traffic environments to support network efficiency and road user safety.

Three operational use cases are presented where computer vision is integrated with existing infrastructure rather than replacing it: the Maioro Dynamic Lane (MDL), the BP Fanshawe Street corridor, and the Dominion Road cyclist safety programme. Across these sites, CV models are used for vehicle, cyclist, and pedestrian detection and classification, trajectory analysis, and compliance monitoring, delivering lane‑level and mode‑specific insights beyond those available from legacy sensors.

The presentation outlines the end‑to‑end AI architecture—from edge capture through to downstream analytics—and discusses key operational considerations including data quality, system reliability, and integration with traffic signal and monitoring systems. Outcomes demonstrate how applied computer vision can be deployed at scale to deliver practical, safety‑focused benefits within complex urban transport networks.