Rethinking Metrics and Benchmarks of Video Anomaly Detection

TL;DR

This paper critically examines current evaluation metrics and benchmarks in Video Anomaly Detection, proposing new methods to address biases, reward early detection, and evaluate scene overfitting, thereby advancing the assessment standards in VAD research.

Contribution

It introduces three novel evaluation metrics and two benchmarks to improve the assessment of VAD models, focusing on bias mitigation, early detection, and overfitting evaluation.

Findings

Prob-AUC/AP reduces annotation bias effects.

LaAP rewards early and accurate detection.

New benchmarks evaluate scene overfitting effectively.

Abstract

Video Anomaly Detection (VAD), which aims to detect anomalies that deviate from expectation, has attracted increasing attention in recent years. Existing advancements in VAD primarily focus on model architectures and training strategies, while devoting insufficient attention to evaluation metrics and benchmarks. In this paper, we rethink VAD evaluation methods through comprehensive analyses, revealing three critical limitations in current practices: 1) existing metrics are significantly influenced by single annotation bias; 2) current metrics fail to reward early detection of anomalies; 3) available benchmarks lack the capability to evaluate scene overfitting of fully/weakly-supervised algorithms. To address these limitations, we propose three novel evaluation methods: first, we establish probabilistic AUC/AP (Prob-AUC/AP) metrics utlizing multi-round annotations to mitigate single annotation bias; second, we develop a Latency-aware Average Precision (LaAP) metric that rewards early and accurate anomaly detection; and finally, we introduce two hard normal benchmarks (UCF-HN, MSAD-HN) with videos specifically designed to evaluate scene overfitting. We report performance comparisons of ten state-of-the-art VAD approaches using our proposed evaluation methods, providing novel perspectives for future VAD model development. We release our data and code in https://github.com/Kamino666/RethinkingVAD.