Custom silicone Face Masks: Vulnerability of Commercial Face Recognition Systems & Presentation Attack Detection
Custom silicone Face Masks: Vulnerability of Commercial Face Recognition Systems & Presentation Attack Detection
复制标题
定制硅胶口罩：商业人脸识别系统的漏洞
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
C. Busch
中科院分区：
文献类型：
--
作者：
Raghavendra Ramachandra;S. Venkatesh;K. Raja;Sushil K. Bhattacharjee;P. Wasnik;S. Marcel;C. Busch
在过去的十年中，人脸识别系统面对不断演变的演示攻击的脆弱性引起了人们的极大兴趣。在本文中，我们使用与真实受试者相对应的定制 3D 硅胶面罩，对商业人脸识别系统 (FRS) 的漏洞分析和演示攻击检测进行了实证研究。为此，我们收集了一个新数据库，其中包含 8 个定制 3D 硅胶面罩以及使用三种不同设备（智能手机）对相应主题的真实演示。使用两种著名的商用 (COTS) FRS（来自 Neurotechnology 的 Verilook 和 Cognitec Face-VACS）有效评估了 3D 面部硅胶面罩的 FRS 的脆弱性。此外，还进行了广泛的实验来评估五种最先进的演示攻击检测（PAD）技术用于检测此类掩码的有效性。提供了有关硅胶面罩 PAD 的重要见解，并讨论了我们实验中实现的准确性。
The vulnerability of face recognition systems towards evolving presentation attacks has drawn significant interest in the last decade. In this paper, we present an empirical study on both vulnerability analysis and presentation attack detection for commercial face recognition systems (FRS) using custom 3D silicone face masks corresponding to real subjects. To this end, a new database is collected consisting of 8 custom 3D silicone masks together with bona fide presentations of the corresponding subjects using three different devices (smartphones). The vulnerability of FRS for 3D face silicone face masks is effectively evaluated using two well-known commercial-off-the-shelf (COTS) FRS (Verilook from Neurotechnology, and the Cognitec Face-VACS). Further, extensive experiments are carried out to evaluate the effectiveness of five state-of-the-art presentation attack detection (PAD) techniques for detecting such masks. Key insights on silicone mask PAD are provided along with a discussion on the accuracy achieved in our experiments.
DOI：
10.1109/cvprw.2017.40
发表时间：
2017-07
期刊：
2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
影响因子：
--
作者：
Akshay Agarwal;Daksha Yadav;Naman Kohli;Richa Singh;Mayank Vatsa;A. Noore
通讯作者：
Akshay Agarwal;Daksha Yadav;Naman Kohli;Richa Singh;Mayank Vatsa;A. Noore