When, Where, and What? A New Dataset for Anomaly Detection in Driving Videos

TL;DR

This paper introduces DoTA, a large-scale egocentric traffic video dataset with annotations and a new evaluation metric, advancing anomaly detection research by providing comprehensive data and benchmarks.

Contribution

It presents the first large-scale traffic anomaly dataset with comprehensive annotations and a novel evaluation metric, enabling more effective anomaly detection in traffic videos.

Findings

STAUC is an effective evaluation metric for VAD.

DoTA is the largest traffic anomaly dataset to date.

Benchmarking shows state-of-the-art methods' performance on DoTA.

Abstract

Video anomaly detection (VAD) has been extensively studied. However, research on egocentric traffic videos with dynamic scenes lacks large-scale benchmark datasets as well as effective evaluation metrics. This paper proposes traffic anomaly detection with a \textit{when-where-what} pipeline to detect, localize, and recognize anomalous events from egocentric videos. We introduce a new dataset called Detection of Traffic Anomaly (DoTA) containing 4,677 videos with temporal, spatial, and categorical annotations. A new spatial-temporal area under curve (STAUC) evaluation metric is proposed and used with DoTA. State-of-the-art methods are benchmarked for two VAD-related tasks.Experimental results show STAUC is an effective VAD metric. To our knowledge, DoTA is the largest traffic anomaly dataset to-date and is the first supporting traffic anomaly studies across when-where-what perspectives. Our code and dataset can be found in: https://github.com/MoonBlvd/Detection-of-Traffic-Anomaly