Sikistmlrrus videoda anomali tespiti

Translated title of the contribution: Anomaly detection in compressed video

Sumeyye Cavas, Muhammet Sebul Beratoglu, Behcet Ugur Toreyin

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)

Abstract

In this paper, an anomaly detection approach has been developed on video compressed in H.265 format. In order to detect anomalies, the motion vectors in the compressed video and the region information of the motion vectors were used. This information was provided as input to the autoencoder model, which is an unsupervised artificial neural network method, and thus the model was trained. The trained model was tested on video data containing anomalies. As output, during the streaming of any video, it is provided to draw a regularity score graph and display the anomaly regions by color. In this paper, we propose an autoencoder based method for anomaly detection in compressed video instead of the original uncompressed video.

Translated title of the contributionAnomaly detection in compressed video
Original languageTurkish
Title of host publicationSIU 2021 - 29th IEEE Conference on Signal Processing and Communications Applications, Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665436496
DOIs
Publication statusPublished - 9 Jun 2021
Event29th IEEE Conference on Signal Processing and Communications Applications, SIU 2021 - Virtual, Istanbul, Turkey
Duration: 9 Jun 202111 Jun 2021

Publication series

NameSIU 2021 - 29th IEEE Conference on Signal Processing and Communications Applications, Proceedings

Conference

Conference29th IEEE Conference on Signal Processing and Communications Applications, SIU 2021
Country/TerritoryTurkey
CityVirtual, Istanbul
Period9/06/2111/06/21

Bibliographical note

Publisher Copyright:
© 2021 IEEE.

Fingerprint

Dive into the research topics of 'Anomaly detection in compressed video'. Together they form a unique fingerprint.

Cite this