An Attribute-based Method for Video Anomaly Detection

Tal Reiss, Yedid Hoshen

2022-12-01Abnormal Event Detection In Video Attribute Density Estimation Video Anomaly Detection Anomaly Detection

Abstract

Video anomaly detection (VAD) identifies suspicious events in videos, which is critical for crime prevention and homeland security. In this paper, we propose a simple but highly effective VAD method that relies on attribute-based representations. The base version of our method represents every object by its velocity and pose, and computes anomaly scores by density estimation. Surprisingly, this simple representation is sufficient to achieve state-of-the-art performance in ShanghaiTech, the most commonly used VAD dataset. Combining our attribute-based representations with an off-the-shelf, pretrained deep representation yields state-of-the-art performance with a $99.1\%, 93.7\%$, and $85.9\%$ AUROC on Ped2, Avenue, and ShanghaiTech, respectively.

Results

Task	Dataset	Metric	Value	Model
Anomaly Detection	UCSD Ped2	AUC	99.1	AI-VAD
Abnormal Event Detection In Video	UCSD Ped2	AUC	99.1	AI-VAD

Related Papers

Missing value imputation with adversarial random forests -- MissARF2025-07-21 Multi-Stage Prompt Inference Attacks on Enterprise LLM Systems2025-07-21 3DKeyAD: High-Resolution 3D Point Cloud Anomaly Detection via Keypoint-Guided Point Clustering2025-07-17 A Semi-Supervised Learning Method for the Identification of Bad Exposures in Large Imaging Surveys2025-07-17 MGFFD-VLM: Multi-Granularity Prompt Learning for Face Forgery Detection with VLM2025-07-16 Non-Adaptive Adversarial Face Generation2025-07-16 A Privacy-Preserving Framework for Advertising Personalization Incorporating Federated Learning and Differential Privacy2025-07-16 Attributes Shape the Embedding Space of Face Recognition Models2025-07-15