A multi-stream deep neural network with late fuzzy fusion for real-world anomaly detection

作者：

Highlights：

• Multi-stream deep learning architecture for anomaly localization in videos.

• Improvement in the conventional MIL classifier for binary classification.

• A late fuzzy fusion technique to improve the anomaly localization scores.

• Classification of road accident and fire related anomalies.

摘要

•Multi-stream deep learning architecture for anomaly localization in videos.•Improvement in the conventional MIL classifier for binary classification.•A late fuzzy fusion technique to improve the anomaly localization scores.•Classification of road accident and fire related anomalies.

论文关键词：Multi-stream networks,Fuzzy fusion,Anomaly detection,Event classification,Multiple instance learning

论文评审过程：Received 14 January 2021, Revised 14 January 2022, Accepted 27 March 2022, Available online 7 April 2022, Version of Record 20 April 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117030