Transformer-Encoder Detector Module: Using Context to Improve Robustness to Adversarial Attacks on Object Detection

Alarmi, F., Kalkan, S. and Pugeault, N. (2021) Transformer-Encoder Detector Module: Using Context to Improve Robustness to Adversarial Attacks on Object Detection. In: 25th International Conference on Pattern Recognition (ICPR2020), Milan, Italy, 10-15 Jan 2021, ISBN 9781728188096 (doi: 10.1109/ICPR48806.2021.9413344)

Abstract

Deep neural network approaches have demonstrated high performance in object recognition (CNN) and detection (Faster-RCNN) tasks, but experiments have shown that such architectures are vulnerable to adversarial attacks (FFF, UAP): low amplitude perturbations, barely perceptible by the human eye, can lead to a drastic reduction in labeling performance. This article proposes a new context module, called \textit{Transformer-Encoder Detector Module}, that can be applied to an object detector to (i) improve the labeling of object instances; and (ii) improve the detector's robustness to adversarial attacks. The proposed model achieves higher mAP, F1 scores and AUC average score of up to 13\% compared to the baseline Faster-RCNN detector, and an mAP score 8 points higher on images subjected to FFF or UAP attacks due to the inclusion of both contextual and visual features extracted from scene and encoded into the model. The result demonstrates that a simple ad-hoc context module can improve the reliability of object detectors significantly.

• Item Type: Conference Proceedings
• Additional Information: Electronic ISBN: 9781728188089.
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Pugeault, Dr Nicolas
• Authors: Alarmi, F., Kalkan, S., and Pugeault, N.
• College/School: College of Science and Engineering > School of Computing Science
• ISBN: 9781728188096
• Copyright Holders: Copyright © 2020 The Authors
• First Published: First published in 25th International Conference on Pattern Recognition (ICPR2020)
• Publisher Policy: Reproduced under a Creative Commons License

Related URLs

• Organisation
• arXiv