Rethinking Metrics and Benchmarks of Video Anomaly Detection

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Zihao, Wu, Xiaoyu, Li, Wenna, Yang, Linlin, Wang, Shengjin
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914125827801088
author Liu, Zihao
Wu, Xiaoyu
Li, Wenna
Yang, Linlin
Wang, Shengjin
author_facet Liu, Zihao
Wu, Xiaoyu
Li, Wenna
Yang, Linlin
Wang, Shengjin
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19022
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Rethinking Metrics and Benchmarks of Video Anomaly Detection
Liu, Zihao
Wu, Xiaoyu
Li, Wenna
Yang, Linlin
Wang, Shengjin
Computer Vision and Pattern Recognition
Artificial Intelligence
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.
title Rethinking Metrics and Benchmarks of Video Anomaly Detection
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2505.19022