Study on evolution mechanism of fateful stampede accident based on graphical evaluation and review technique

Hu, Mingli; Fang, Zhigeng; Liu, Sifeng; Chen, Wei; Yang, Banghua

doi:10.1109/icsmc.2010.5641989

Cited by 1 publication

(1 citation statement)

References 4 publications

(7 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We design our skipping frames pseudo anomaly synthesizer based on the intuition that detecting fast or suddenly changing motions is significantly important and closely related to anomalous events. For instance, people may run unusually if there are life-threatening situations nearby, such as natural disasters [65,66] or fires [67,68]. Moreover, robberies and fights may also involve sudden strong motions.…”

Section: Skip Framesmentioning

confidence: 99%

PseudoBound: Limiting the anomaly reconstruction capability of one-class classifiers using pseudo anomalies

Astrid,

Zaheer,

Lee

2023

Preprint

View full text Add to dashboard Cite

Due to the rarity of anomalous events, video anomaly detection is typically approached as one-class classification (OCC) problem. Typically in OCC, an autoencoder (AE) is trained to reconstruct the normal only training data with the expectation that, in test time, it can poorly reconstruct the anomalous data. However, previous studies have shown that, even trained with only normal data, AEs can often reconstruct anomalous data as well, resulting in a decreased performance. To mitigate this problem, we propose to limit the anomaly reconstruction capability of AEs by incorporating pseudo anomalies during the training of an AE. Extensive experiments using five types of pseudo anomalies show the robustness of our training mechanism towards any kind of pseudo anomaly. Moreover, we demonstrate the effectiveness of our proposed pseudo anomaly based training approach against several existing state-of-the-art (SOTA) methods on three benchmark video anomaly datasets, outperforming all the other reconstruction-based approaches in two datasets and showing the second best performance in the other dataset.

show abstract