Trajectory-Based Anomaly Detection in Highway Traffic Surveillance Videos Using Unsupervised Learning Techniques
G.W.P.R.R. Wijesinghe, W.S.T. Sandaruwan, Nandith Sajith, M. N. M. Aashiq · 2024
An effective novel abnormal event detection approach is proposed in traffic monitoring systems based on trajectory data analysis to enhance the accuracy and efficiency of anomaly identification. The proposed method introduces a novel data manipulation algorithm for the effectively distinguishing of abnormal trajectory data from normal trajectory data. The unsupervised approach used for effectively recognize various traffic incidents, such as abrupt stoppages, vehicle breakdowns, and accidents. Considering a dataset of high-definition videos, the results demonstrate a significant increase in the rate of detection and a considerable reduction of false positives compared to the traditional methods. This innovative approach not only streamlines the process of real-time anomaly detection but also provides a scalable solution adaptable to various traffic conditions and environments. These results highlight the promise of this approach in further contributing to improve road safety and optimization of traffic management systems, thus paving the way for future research and applications in intelligent transportation systems. Notably, our model achieved an F1 score of 0.9230, underscoring its robustness and reliability in real-world applications.