Automatic Incident Classification for Large-scale Traffic Data

  • NGAN, Henry Y T (PI)

Project: Research project

Project Details


This proposal of automatic incident classification (AIC) for large-scale traffic data is mainly to contribute to the development of traffic monitoring and managing for Traffic Control and Surveillance System (TCSS). An effective AIC method enhances the response time to traffic incidents and reduces massive human operators in operation. It should be capable to detect and classify a traffic incident from massive data and prevent the result from any corruption of data errors. Over decades, most previous research has mainly focused on the automatic incident detection (AID), with or without incident only. This means no detailed classification of traffic incidents was given. Also, they rarely dealed with data errors embedded on the dataset. Therefore, it is increasing a demand to develop a technology to improve the data quality and strengthen the discriminative power to detect and classify traffic incident. First, it is not straightforward to separate traffic incident and data errors. This proposal offers a new perspective to treat this problem from outlier detection and classification. Outliers are those inconsistent to the majority of data. Data errors and traffic incidents embedded in traffic datasets are outliers while others are regarded as inliers. Hence, this project will develop the AIC based on a framework of outlier detection and classification method. Second, traffic incident classification requires more advanced techniques and strong support of an accurate traffic incident detection result. Therefore, in this proposal, our AIC would firstly screen outliers (traffic incidents and data errors) and inliers (incident-free events) from massive traffic data by an outlier detection method with high detection rate. Then, our AIC will perform a finer outlier classification of detected anomalies for 7 outlier types, in which two types belong to data errors and five types belong to traffic incidents. This would contribute to the state-of-the-art TCSS with automatic and accurate traffic incident detection, and fine classification of traffic incidents.
Effective start/end date1/10/1431/03/17

UN Sustainable Development Goals

In 2015, UN member states agreed to 17 global Sustainable Development Goals (SDGs) to end poverty, protect the planet and ensure prosperity for all. This project contributes towards the following SDG(s):

  • SDG 3 - Good Health and Well-being
  • SDG 11 - Sustainable Cities and Communities


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.