DEVELOPING A REAL-WORLD TESTBED FOR AUTOMATIC INCIDENT DETECTION SYSTEMS

Abstract Presented is a real-world testbed for Automatic Incident Detection (AID) systems that consists of an online data warehouse storing a month of traffic video, the corresponding traffic data and an operator log of incident start/end times. A proof-of-concept field evaluation is conducted whereby the testbed is used to calibrate, and then analyze the performance of four AID systems: California Algorithm 8, McMaster Algorithm, GAID Algorithm, and Citlog-VisioPAD. The pilot evaluation shows considerable advantages of the testbed in its ability to analyse the performance of the AID systems.