Anomaly Behavior Detection Based on Deep Learning in an IoT Environment

Anqi Fu, Jian Li, Gregor Schiele, Alessandra Rizzardi · 2025

Read Anomaly Behavior Detection Based on Deep Learning in an IoT Environment by Anqi Fu, Jian Li, Gregor Schiele, Alessandra Rizzardi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12737123.txt.

Description

Smart city cameras pour out video, and spotting abnormal behavior in that flood is a core surveillance task. Weakly supervised detectors struggle: abnormal clips are scarce, normal and abnormal segments look alike, and time structure is under-modeled. The authors add temporal structural attention, armed with causal masks and decay weights so the model respects time order and never peeks at the future. They also add contrastive learning with positive and negative offsets to pull abnormal segments apart in the latent space.

On several public video anomaly data sets the method beat mainstream models, reaching an AUC of 98.1 percent, accuracy of 96.1 percent, and an F1 score of 94.5 percent. The authors present it as a more reliable anomaly detector for Internet of Things surveillance and public safety.

Catalog record

CREATORAnqi Fu, Jian Li, Gregor Schiele, Alessandra Rizzardi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE83 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC12737123

Browse the full library · About the Tower of Babel