phung-robustanomaly-2019.pdf (3.86 MB)
Robust anomaly detection in videos using multilevel representations
conference contribution
posted on 2019-01-01, 00:00 authored by Hung Vu, Dinh Nguyen Tu, Le Trung, Wei LuoWei Luo, Phung DinhDetecting anomalies in surveillance videos has long been an important but unsolved problem. In particular, many existing solutions are overly sensitive to (often ephemeral) visual artifacts in the raw video data, resulting in false positives and fragmented detection regions. To overcome such sensitivity and to capture true anomalies with semantic significance, one natural idea is to seek validation from abstract representations of the videos. This paper introduces a framework of robust anomaly detection using multilevel representations of both intensity and motion data. The framework consists of three main components: 1) representation learning using Denoising Autoencoders, 2) level-wise representation generation using Conditional Generative Adversarial Networks, and 3) consolidating anomalous regions detected at each representation level. Our proposed multilevel detector shows a significant improvement in pixel-level Equal Error Rate, namely 11.35%, 12.32% and 4.31% improvement in UCSD Ped 1, UCSD Ped 2 and Avenue datasets respectively. In addition, the model allowed us to detect mislabeled anomalies in the UCDS Ped 1.
History
Event
AAAI Conference on Artificial Intelligence., Innovative Applications of Artificial Intelligence Conference and AAAI Symposium on Educational Advances in Artificial Intelligence. Combined Conference (2019 : 33rd, 31st & 9th : Honolulu, Hawaii)Volume
33Issue
1Pagination
5216 - 5223Publisher
AAAI PressLocation
Honolulu, HawaiiPlace of publication
[Honolulu, Hawaii]Publisher DOI
Link to full text
Start date
2019-01-27End date
2019-02-01eISSN
2374-3468ISBN-13
9781577358091Language
engPublication classification
E1 Full written paper - refereedTitle of proceedings
Proceedings of the Combined Conferences : 33rd AAAI Conference on Artificial Intelligence / 31st Innovative Applications of Artificial Intelligence Conference / 9th AAAI Symposium on Educational Advances in Artificial IntelligenceUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC