SavedAdd your journal

Journals / Iğdır Üniversitesi Fen Bilimleri Enstitüsü Dergisi / 2024 / Cilt: 14 Sayı: 4

The Impact of Adversarial Attacks on Deep Learning-based Face Anti-Spoofing Systems

Çekişmeli Saldırıların Derin Öğrenme Tabanlı Yüz Sahteciliği Önleme Sistemlerine Etkisi

Ersin Kılıç(ArkSigner Yazılım ve Donanım San. A.Ş.)0000-0002-0924-9246
Ömür Şahin(ERCİYES ÜNİVERSİTESİ)0000-0003-1213-7445
Ahmet Nusret Toprak(ERCİYES ÜNİVERSİTESİ)0000-0003-4841-9508
Pages
1397–1407
Publication Date
2024-12-01
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Experimental results demonstrate that the DGUA-FAS model misclassifies all attacked images.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Face recognition and verification systems are widely employed in various applications, from smartphones and security systems to payment and remote healthcare services, demonstrating high accuracy.
📊
Key Findings & Data:These models are expected to accurately detect diverse forgery techniques and be resilient to adversarial attacks on input images.
💡
Conclusion & Impact:Experimental results demonstrate that the DGUA-FAS model misclassifies all attacked images.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Yüz sahteciliği önleme · Çekişmeli saldırı · Derin öğrenme

Abstract

Face recognition and verification systems are widely employed in various applications, from smartphones and security systems to payment and remote healthcare services, demonstrating high accuracy. However, a common method to spoof these systems involves presenting a different person's face using tools such as masks, printouts, or monitors instead of the actual user's face. In recent years, similar to advancements in other computer vision tasks, deep neural networks have been developed to effectively combat image forgery in face recognition systems. These models are expected to accurately detect diverse forgery techniques and be resilient to adversarial attacks on input images. This study investigates the robustness of DGUA-FAS, a state-of-the-art image forgery detection model, against adversarial attacks. Adversarial examples are generated using a black-box adversarial attack generation method based on MIO algorithm, to mislead the DGUA-FAS model. Experimental results demonstrate that the DGUA-FAS model misclassifies all attacked images. The findings highlight the necessity for developing face forgery detection models that are more resilient to adversarial attacks.

Özet

Günümüzde yüz tanıma ve doğrulama sistemleri akıllı telefonlardan güvenlik sistemlerine, ödeme uygulamalarından uzaktan sağlık uygulamalarına kadar birçok alanda yüksek doğrulukla kullanılmaktadır. Yüz tanıma sistemlerini yanıltmanın en yaygın yolu kişinin sisteme kendi yüzü yerine maske, yazıcı çıktısı ya da monitör gibi araçlarla başka birinin yüzünü sunmasıdır. Son yıllarda birçok bilgisayarla görme uygulamasında olduğu gibi yüz tanıma sistemlerinde görüntü sahteciliğini önleme amacıyla da oldukça başarılı derin sinir ağı modelleri geliştirilmiştir. Bu modellerin farklı sahtecilik yöntemlerini hassas bir şekilde tespit edebilmesi ve girdi görüntülere uygulanacak saldırılara karşı dayanıklı olması beklenmektedir. Bu çalışmada güncel ve başarılı bir görüntü sahteciliği tespit modeli olan DGUA-FAS'ın çekişmeli saldırılara karşı dayanıklılığı araştırılmaktadır. Bu amaçla MIO algoritmasına dayalı kara kutu çekilmeli saldırı üretme yöntemi ile DGUA-FAS modelini yanıltmak amacıyla çekişmeli örnek görüntüler üretilmektedir. Deneysel çalışmalar, DGUA-FAS modelinin saldırı uygulanan tüm görüntüleri hatalı şekilde sınıflandırdığını göstermektedir. Elde edilen sonuçlar, yüz sahteciliği tespit modellerinin çekişmeli saldırılara karşı daha dayanıklı hale gelmesi gerektiğini göstermektedir.

🔖

Cite This Article

Kılıç, E., Şahin, Ö., Toprak, A. N. (2024). The Impact of Adversarial Attacks on Deep Learning-based Face Anti-Spoofing Systems. Iğdır Üniversitesi Fen Bilimleri Enstitüsü Dergisi, Cilt: 14 Sayı: 4, 1397–1407. https://doi.org/10.21597/jist.1533471

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (20)

  1. Akhtar, N., Mian, A., Kardan, N., & Shah, M. (2021). Advances in adversarial attacks and defenses in computer vision: A survey. IEEE Access, 9, 155161-155196.
  2. Arcuri, A. (2018). Test suite generation with the Many Independent Objective (MIO) algorithm. Information and Software Technology, 104, 195-206.
  3. Atoum, Y., Liu, Y., Jourabloo, A., & Liu, X. (2017, October). Face anti-spoofing using patch and depth-based CNNs. In 2017 IEEE international joint conference on biometrics (IJCB) (pp. 319-328). IEEE.
  4. Bartlett, A., Liem, C. C., & Panichella, A. (2023, May). On the Strengths of Pure Evolutionary Algorithms in Generating Adversarial Examples. In 2023 IEEE/ACM International Workshop on Search-Based and Fuzz Testing (SBFT) (pp. 1-8). IEEE.
  5. Dong, Y., Liao, F., Pang, T., Hu, X., & Zhu, J. (2017). Discovering adversarial examples with momentum. arXiv preprint arXiv:1710.06081, 5.
  6. Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014). Explaining and harnessing adversarial examples. arXiv preprint arXiv:1412.6572.
  7. Hong, Z. W., Lin, Y. C., Liu, H. T., Yeh, Y. R., & Chen, C. S. (2023, October). Domain-Generalized Face Anti-Spoofing with Unknown Attacks. In 2023 IEEE International Conference on Image Processing (ICIP) (pp. 820-824). IEEE.
  8. Jourabloo, A., Liu, Y., & Liu, X. (2018). Face de-spoofing: Anti-spoofing via noise modeling. In Proceedings of the European conference on computer vision (ECCV) (pp. 290-306).
  9. Kurakin, A., Goodfellow, I. J., & Bengio, S. (2018). Adversarial examples in the physical world. In Artificial intelligence safety and security (pp. 99-112). Chapman and Hall/CRC.
  10. Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2017). Towards deep learning models resistant to adversarial attacks. arXiv preprint arXiv:1706.06083.
  11. Papernot, N., McDaniel, P., Jha, S., Fredrikson, M., Celik, Z. B., & Swami, A. (2016, March). The limitations of deep learning in adversarial settings. In 2016 IEEE European symposium on security and privacy (EuroS&P) (pp. 372-387). IEEE.
  12. Sahin, O. (2024). Generation of black-box adversarial attacks using many independent objective-based algorithm for testing the robustness of deep neural networks. Applied Soft Computing, 164, 111969.
  13. Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013). Intriguing properties of neural networks. arXiv preprint arXiv:1312.6199.
  14. Taşyürek, M., & Gül, E. (2023). Nesne Tespitinde En Uygun Modelin Seçimi İçin Görüntüler Üzerinde Evrişimli Sinir Ağları ile Çekişmeli Saldırı Tespiti. Journal of the Institute of Science and Technology, 13(4), 2353-2363.
  15. Wang, Z., Yu, Z., Zhao, C., Zhu, X., Qin, Y., Zhou, Q.,. & Lei, Z. (2020). Deep spatial gradient and temporal depth learning for face anti-spoofing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5042-5051).
  16. Yang, J., Lei, Z., & Li, S. Z. (2014). Learn convolutional neural network for face anti-spoofing. arXiv preprint arXiv:1408.5601.
  17. Yu, Z., Zhao, C., Wang, Z., Qin, Y., Su, Z., Li, X.,. & Zhao, G. (2020). Searching central difference convolutional networks for face anti-spoofing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5295-5305).
  18. Yu, Z., Qin, Y., Li, X., Zhao, C., Lei, Z., & Zhao, G. (2022). Deep learning for face anti-spoofing: A survey. IEEE transactions on pattern analysis and machine intelligence, 45(5), 5609-5631.
  19. Wang, Z., Yu, Z., Zhao, C., Zhu, X., Qin, Y., Zhou, Q.,. & Lei, Z. (2020). Deep spatial gradient and temporal depth learning for face anti-spoofing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5042-5051).
  20. Yu, Z., Zhao, C., Wang, Z., Qin, Y., Su, Z., Li, X.,. & Zhao, G. (2020). Searching central difference convolutional networks for face anti-spoofing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 5295-5305).