Layer-wise External Attention for Efficient Deep Anomaly Detection

Tokihisa Hayakawa, Keiichi Nakanishi, Ryoya Katafuchi, Terumasa Tokunaga

2023

Abstract

Recently, the visual attention mechanism has become a promising way to improve the performance of Convolutional Neural Networks (CNNs) for many applications. In this paper, we propose a Layer-wise External Attention mechanism for efficient image anomaly detection. The core idea is the integration of unsupervised and supervised anomaly detectors via the visual attention mechanism. Our strategy is as follows: (i) prior knowledge about anomalies is represented as an anomaly map generated by the pre-trained network; (ii) the anomaly map is translated to an attention map via an external network. (iii) the attention map is then incorporated into intermediate layers of the anomaly detection network via visual attention. Notably, the proposed method can be applied to any CNN model in an end-to-end training manner. We also propose an example of a network with Layer-wise External Attention called Layer-wise External Attention Network (LEA-Net). Through extensive experiments using real-world datasets, we demonstrate that Layer-wise External Attention consistently boosts the anomaly detection performances of an existing CNN model, even on small and unbalanced data. Moreover, we show that Layer-wise External Attention works well with Self-Attention Networks.

Paper Citation

in Harvard Style

Hayakawa T., Nakanishi K., Katafuchi R. and Tokunaga T. (2023). Layer-wise External Attention for Efficient Deep Anomaly Detection. In Proceedings of the 3rd International Conference on Image Processing and Vision Engineering - Volume 1: IMPROVE, ISBN 978-989-758-642-2, SciTePress, pages 100-110. DOI: 10.5220/0011856800003497

in Bibtex Style

@conference{improve23,
author={Tokihisa Hayakawa and Keiichi Nakanishi and Ryoya Katafuchi and Terumasa Tokunaga},
title={Layer-wise External Attention for Efficient Deep Anomaly Detection},
booktitle={Proceedings of the 3rd International Conference on Image Processing and Vision Engineering - Volume 1: IMPROVE,},
year={2023},
pages={100-110},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011856800003497},
isbn={978-989-758-642-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Conference on Image Processing and Vision Engineering - Volume 1: IMPROVE,
TI - Layer-wise External Attention for Efficient Deep Anomaly Detection
SN - 978-989-758-642-2
AU - Hayakawa T.
AU - Nakanishi K.
AU - Katafuchi R.
AU - Tokunaga T.
PY - 2023
SP - 100
EP - 110
DO - 10.5220/0011856800003497
PB - SciTePress