arXiv:2503.04504v4 Announce Type: replace 
Abstract: Video anomaly detection (VAD) is crucial for video analysis and surveillance in computer vision. However, existing VAD models rely on learned normal patterns, which makes them difficult to apply to diverse environments. Consequently, users should retrain models or develop separate AI models for new environments, which requires expertise in machine learning, high-performance hardware, and extensive data collection, limiting the practical usability of VAD. To address these challenges, this study proposes customizable video anomaly detection (C-VAD) technique and the AnyAnomaly model. C-VAD considers user-defined text as an abnormal event and detects frames containing a specified event in a video. We effectively implemented AnyAnomaly using a context-aware visual question answering without fine-tuning the large vision language model. To validate the effectiveness of the proposed model, we constructed C-VAD datasets and demonstrated the superiority of AnyAnomaly. Furthermore, our approach showed competitive results on VAD benchmarks, achieving state-of-the-art performance on UBnormal and UCF-Crime and surpassing other methods in generalization across all datasets. Our code is available online at github.com/SkiddieAhn/Paper-AnyAnomaly.

يقدم نموذج AnyAnomaly تقنية الكشف عن الشذوذ في الفيديو القابلة للتخصيص (C-VAD) التي تتيح للمستخدمين تعريف الأحداث الشاذة من خلال النص، مما يعزز قابلية تطبيق تحليل الفيديو في بيئات متنوعة دون الحاجة إلى إعادة تدريب مكثف.

El modelo AnyAnomaly presenta una técnica de detección de anomalías en video personalizable (C-VAD) que permite a los usuarios definir eventos anormales a través de texto, mejorando la aplicabilidad del análisis de video en diversos entornos sin necesidad de un extenso reentrenamiento.

Le modèle AnyAnomaly introduit une technique de détection d'anomalies vidéo personnalisable (C-VAD) permettant aux utilisateurs de définir des événements anormaux par le biais de textes, améliorant ainsi l'applicabilité de l'analyse vidéo dans divers environnements sans nécessiter de réentraînement extensif.

The AnyAnomaly model introduces a customizable video anomaly detection (C-VAD) technique that allows users to define abnormal events through text, enhancing the applicability of video analysis in diverse environments without the need for extensive retraining.

AnyAnomaly: Zero-Shot Customizable Video Anomaly Detection with LVLM

Was this article worth reading? Share it

LucidQuery AI

Videotok

Videolulu

Ready to build your own newsroom?