Using Feature Alignment Can Improve Clean Average Precision and Adversarial Robustness in Object Detection

TL;DR

This paper introduces feature alignment modules to enhance both clean average precision and adversarial robustness in object detection, addressing the trade-off caused by adversarial training.

Contribution

It proposes knowledge-distilled and self-supervised feature alignment modules that improve detection performance and robustness simultaneously.

Findings

Feature alignment improves clean AP in object detection.

The modules enhance robustness against adversarial attacks.

Experimental results on PASCAL VOC and MS-COCO validate effectiveness.

Abstract

The 2D object detection in clean images has been a well studied topic, but its vulnerability against adversarial attack is still worrying. Existing work has improved robustness of object detectors by adversarial training, at the same time, the average precision (AP) on clean images drops significantly. In this paper, we propose that using feature alignment of intermediate layer can improve clean AP and robustness in object detection. Further, on the basis of adversarial training, we present two feature alignment modules: Knowledge-Distilled Feature Alignment (KDFA) module and Self-Supervised Feature Alignment (SSFA) module, which can guide the network to generate more effective features. We conduct extensive experiments on PASCAL VOC and MS-COCO datasets to verify the effectiveness of our proposed approach. The code of our experiments is available at https://github.com/grispeut/Feature-Alignment.git.