世界银行：2026年在低收入城市环境中使用计算机视觉分析测量道路安全指标：亚的斯亚贝巴案例研究报告（英文版）

he report’s objective is to present and validate an open-source, practical computer vision tool—the Vehicle Incidence Detection Algorithm (VIDA). This tool can continuously generate traffic safety data (e.g., vehicle flows, near-misses, and risky behaviors) at high resolution from closed-circuit television (CCTV) footage in low-income urban contexts, as piloted in Addis Ababa. Such data can then be analyzed by authorities to diagnose risks, prioritize investments, and verify safety benefits o