This paper investigates the performance of diffusion models for video anomaly detection (VAD) within the most challenging but also the most operational scenario in which the data annotations are not used. As being sparse, diverse, contextual, and often ambiguous, detecting abnormal events precisely is a very ambitious task. To this end, we rely only on the information-rich spatio-temporal data, and the reconstruction power of the diffusion models such that a high reconstruction error is utilized to decide the abnormality. Experiments performed on two large-scale video anomaly detection datasets demonstrate the consistent improvement of the proposed method over the state-of-the-art generative models while in some cases our method achieves better scores than the more complex models. This is the first study using a diffusion model and examining its parameters' influence to present guidance for VAD in surveillance scenarios.

Exploring Diffusion Models for Unsupervised Video Anomaly Detection / Osman Tur, Anil; Dall'Asen, Nicola; Beyan, Cigdem; Ricci, Elisa. - (2023), pp. 2540-2544. (Intervento presentato al convegno 30th IEEE International Conference on Image Processing, ICIP 2023 tenutosi a Kuala Lumpur, Malaysia nel 8th - 11th October, 2023) [10.1109/ICIP49359.2023.10222594].

Exploring Diffusion Models for Unsupervised Video Anomaly Detection

Osman Tur, Anil;Dall'Asen, Nicola;Beyan, Cigdem;Ricci, Elisa
2023-01-01

Abstract

This paper investigates the performance of diffusion models for video anomaly detection (VAD) within the most challenging but also the most operational scenario in which the data annotations are not used. As being sparse, diverse, contextual, and often ambiguous, detecting abnormal events precisely is a very ambitious task. To this end, we rely only on the information-rich spatio-temporal data, and the reconstruction power of the diffusion models such that a high reconstruction error is utilized to decide the abnormality. Experiments performed on two large-scale video anomaly detection datasets demonstrate the consistent improvement of the proposed method over the state-of-the-art generative models while in some cases our method achieves better scores than the more complex models. This is the first study using a diffusion model and examining its parameters' influence to present guidance for VAD in surveillance scenarios.
2023
Proceedings of the IEEE International Conference on Image Processing (ICIP 2023)
Piscataway, NJ USA
IEEE Computer Society
978-1-7281-9835-4
978-1-7281-9836-1
Osman Tur, Anil; Dall'Asen, Nicola; Beyan, Cigdem; Ricci, Elisa
Exploring Diffusion Models for Unsupervised Video Anomaly Detection / Osman Tur, Anil; Dall'Asen, Nicola; Beyan, Cigdem; Ricci, Elisa. - (2023), pp. 2540-2544. (Intervento presentato al convegno 30th IEEE International Conference on Image Processing, ICIP 2023 tenutosi a Kuala Lumpur, Malaysia nel 8th - 11th October, 2023) [10.1109/ICIP49359.2023.10222594].
File in questo prodotto:
File Dimensione Formato  
IC25_Exploring Diffusion Models for Unsupervised Video Anomaly Detection.pdf

accesso aperto

Tipologia: Post-print referato (Refereed author’s manuscript)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 3.18 MB
Formato Adobe PDF
3.18 MB Adobe PDF Visualizza/Apri
Exploring_Diffusion_Models_for_Unsupervised_Video_Anomaly_Detection.pdf

Solo gestori archivio

Tipologia: Versione editoriale (Publisher’s layout)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 3.88 MB
Formato Adobe PDF
3.88 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11572/387313
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 15
  • ???jsp.display-item.citation.isi??? 7
  • OpenAlex ND
social impact