当前位置: X-MOL 学术Vis. Comput. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Adversarial defenses for object detectors based on Gabor convolutional layers
The Visual Computer ( IF 3.0 ) Pub Date : 2021-07-24 , DOI: 10.1007/s00371-021-02256-6
Abdollah Amirkhani 1 , Mohammad Parsa Karimi 1
Affiliation  

Despite their many advantages and positive features, the deep neural networks are extremely vulnerable against adversarial attacks. This drawback has substantially reduced the adversarial accuracy of the visual object detectors. To make these object detectors robust to adversarial attacks, a new Gabor filter-based method has been proposed in this paper. This method has then been applied on the YOLOv3 with different backbones, the SSD with different input sizes and on the FRCNN; and thus, six robust object detector models have been presented. In order to evaluate the efficacy of the models, they have been subjected to adversarial training via three types of targeted attacks (TOG-fabrication, TOG-vanishing, and TOG-mislabeling) and three types of untargeted random attacks (DAG, RAP, and UEA). The best average accuracy (49.6%) was achieved by the YOLOv3-d model, and for the PASCAL VOC dataset. This is far superior to the best performance and accuracy and obtained in previous works (25.4%). Empirical results show that, while the presented approach improves the adversarial accuracy of the object detector models, it does not affect the performance of these models on clean data.



中文翻译:

基于 Gabor 卷积层的目标检测器对抗防御

尽管具有许多优点和积极特征,但深度神经网络极易受到对抗性攻击。这个缺点大大降低了视觉对象检测器的对抗精度。为了使这些对象检测器对对抗性攻击具有鲁棒性,本文提出了一种新的基于 Gabor 滤波器的方法。该方法随后被应用于具有不同主干的 YOLOv3、具有不同输入大小的 SSD 和 FRCNN;因此,已经提出了六个鲁棒的物体检测器模型。为了评估模型的有效性,他们通过三种类型的有针对性的攻击(TOG-fabrication、TOG-vanishing 和 TOG-mislabeling)和三种类型的无目标随机攻击(DAG、RAP 和欧洲联盟)。最佳平均准确率 (49. 6%) 是由 YOLOv3-d 模型和 PASCAL VOC 数据集实现的。这远远优于之前工作中获得的最佳性能和准确度(25.4%)。实证结果表明,虽然所提出的方法提高了对象检测器模型的对抗精度,但它不会影响这些模型在干净数据上的性能。

更新日期:2021-07-24
down
wechat
bug