Abstract
Face recognition (FR) is one of the most successful image analysis and understanding applications, which has recently received significant attentions. However, despite the remarkable progress in face recognition-related technologies, dependably recognizing surveillance faces is still a big challenge due to different data distributions between commonly used faces and surveillance faces. Although collecting labeled surveillance faces could be direct and helpful, it is infeasible due to privacy and labor cost. In comparison, it is practical to use the unsupervised domain adaptation (UDA) method to regularize the learning of surveillance face representations by utilizing small-scale unlabeled data. In this paper, a joint pixel-level and feature-level UDA framework has been proposed for tacking the above problems. Specifically, we introduce a training method of domain adversarial to align the domain gap in feature space. Additionally, we propose to use a CycleGAN-based image style transformer to capture pixel-level domain shifts. Moreover, we adopt feature identity-consistency loss on the original CycleGAN to alleviate the identity attribute shift caused by unstable image style transfer. The experimental results on the self-built SFace-450 dataset demonstrate that our method consistently outperforms the state-of-the-art FR and UDA methods.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Wang, M., Deng, W.: Deep visual domain adaptation: a survey. Neurocomputing 312, 135–153 (2018)
Shi, Y., Anil, K.J.: Boosting unconstrained face recognition with auxiliary unlabeled data. In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 2789–2798 (2021)
Dong, Y., Zhen, L., Liao, S., Li, S.Z.: Learning face representation from scratch. arXiv preprint arXiv:1411.7923 (2014)
Cao, Q., et al.: Vggface2: a dataset for recognising faces across pose and age. In: 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018), pp. 67–74. IEEE (2018)
Hong, S., Im, W., Ryu, J., Yang, H.S.: SSPP-DAN: Deep domain adaptation network for face recognition with single sample per person. In: 2017 IEEE International Conference on Image Processing (ICIP) (2017)
Tzeng, E., et al.: Deep domain confusion: Maximizing for domain invariance. arXiv preprint arXiv:1412.3474 (2014)
Long, M., et al.: Learning transferable features with deep adaptation networks. In: ICML, pp. 97–105 (2015)
Ganin, Y., Victor L.: Unsupervised domain adaptation by backpropagation. In: ICML, pp. 1180–1189 (2015)
Tzeng, E., Hoffman, J., Saenko, K., Darrell, T.: Adversarial discriminative domain adaptation. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 7167–7176 (2017)
Hoffman, J., et al.: CyCADA: Cycle-consistent adversarial domain adaptation. In: ICML, pp. 1989–1998 (2018)
Zhu, J.Y., et al.: Unpaired image-to-image translation using cycle-consistent adversarial networks. In: 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pp. 2223–2232 (2017)
Liu, M.Y., Thomas, B., Jan K.: Unsupervised image-to-image translation networks. In: Advances in Neural Information Processing Systems, pp. 701–709 (2017)
Taigman, Y., et al.: Deepface: Closing the gap to human-level performance in face verification. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1701–1708 (2014)
Sun, Y., et al.: Deep learning face representation by joint identification-verification. In: Advances in Neural Information Processing Systems, pp. 1988–1996 (2014)
Schroff, F., Dmitry, K., James, P.: Facenet: A unified embedding for face recognition and clustering. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 815–823 (2015)
Wen, Y., Zhang, K., Li, Z., Qiao, Y.: A discriminative feature learning approach for deep face recognition. In: Leibe, B., Matas, J., Sebe, N., Welling, M. (eds.) ECCV 2016. LNCS, vol. 9911, pp. 499–515. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-46478-7_31
Ranjan, R., Castillo, C.D., Chellppa, R..: L2-constrained softmax loss for discriminative face verification. arXiv preprint arXiv:1703.09507 (2017)
Liu, W., et al.: Sphereface: Deep hypersphere embedding for face recognition. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 6738–6746 (2017)
Wang, H., et al.: Cosface: Large margin cosine loss for deep face recognition. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 5265–5274 (2018)
Deng, J., et al.: Arcface: Additive angular margin loss for deep face recognition. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4685–4694 (2019)
Huang, G.B., et al.: Labeled faces in the wild: a database forstudying face recognition in unconstrained environments. In: Workshop on faces in’Real-Life’Images: detection, alignment, and recognition (2008)
Whitelam, C., et al.: Iarpa janus benchmark-b face dataset. In: 2017 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 592–600 (2017)
Isola, P., Zhu, J.Y., Zhou, T., Efros, A.A.: Image-to-image translation with conditional adversarial networks. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 5967–5976 (2017)
Mirza, M., Simon O.: Conditional generative adversarial nets. arXiv preprint arXiv: 1411.1784 (2014)
Zhang, K., et al.: Joint face detection and alignment using multitask cascaded convolutional networks. IEEE Signal Process. Lett. 23(10), 1499–1503 (2016)
Ye, M., et al.: Deep learning for person re-identification: a survey and outlook. IEEE TPAMI 44(6), 2872–2893 (2022)
Acknowledgement
This work is supported by Industry-University Cooperation Project of Fujian Science and Technology Department (No. 2021H6035), and the Science and Technology Planning Project of Fujian Province (No. 2021J011182, 2020H0023, 2020Y9064), and the Joint Funds of 5th Round of Health and Education Research Program of Fujian Province (No. 2019-WJ-41).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Zhu, H., Yin, H., Xia, D., Wang, DH., Liu, X., Zhu, S. (2022). Joint Pixel-Level and Feature-Level Unsupervised Domain Adaptation for Surveillance Face Recognition. In: Yu, S., et al. Pattern Recognition and Computer Vision. PRCV 2022. Lecture Notes in Computer Science, vol 13536. Springer, Cham. https://doi.org/10.1007/978-3-031-18913-5_36
Download citation
DOI: https://doi.org/10.1007/978-3-031-18913-5_36
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-18912-8
Online ISBN: 978-3-031-18913-5
eBook Packages: Computer ScienceComputer Science (R0)