Audio-Visual Dataset and Method for Anomaly Detection in Traffic Videos

TL;DR

This paper presents MAVAD, a novel audio-visual dataset for traffic anomaly detection, and introduces AVACA, a cross-attention based method that leverages both audio and visual data to improve detection accuracy.

Contribution

The paper introduces the first real-world audio-visual dataset MAVAD and proposes AVACA, a new method that combines audio and visual features for enhanced traffic anomaly detection.

Findings

Audio improves detection performance by up to 5.2%.

Image anonymization causes only a 1.7% decrease in performance.

MAVAD covers diverse weather and illumination conditions.

Abstract

We introduce the first audio-visual dataset for traffic anomaly detection taken from real-world scenes, called MAVAD, with a diverse range of weather and illumination conditions. In addition, we propose a novel method named AVACA that combines visual and audio features extracted from video sequences by means of cross-attention to detect anomalies. We demonstrate that the addition of audio improves the performance of AVACA by up to 5.2%. We also evaluate the impact of image anonymization, showing only a minor decrease in performance averaging at 1.7%.