人臉辨識一直以來都是相當熱門的影像處理的項目，尤其是基於深度學習方法的影
像處理方法。各式各樣不同的技術在過去數年內不斷的被提出且取得了長足的進步。為
了減少人工標註，越來越多使用未標註資料訓練人臉辨識模型的技術被開發，然而這些
方法很少考慮到在現實世界中取得未標註資料的狀況。現實狀況下取得的未標註資料可
能會包含大量的不同個體，但對同一個體卻缺乏多樣性的訓練資料，這樣對同一個體缺
乏多樣性訓練樣本的資料集相對於同一個體具有多樣性的訓練樣本的資料集又被稱作淺
資料集。在本篇論文中，我們提出模擬實例與分離表示損失函數兩個創新方法可以簡單
又有效的在使用淺資料集狀況下訓練深度學習模型。分離表示損失函數藉由分離未標註
資料與標註資料的餘弦相似度提升不同類別資料的差距。模擬實例藉由生成模擬的實例
特徵幫助分類更加準確。藉由我們提出的兩個新技術，實驗結果顯示可以有效的在使用
淺資料集的狀況下依然訓練出相當優秀的深度學習模型。

Face recognition is always a popular task. Many deep learning methods are proposed
and made substantial progress in recent years. More and More methods are proposed to
exploit unlabeled data for training face recogniton models. However, few of them considered the real­world scenarios of collected unlabeled data from the wild. In this paper,
we proposed two novel methods, Segregate­representation loss and SimInstance, which
are simple but effective against shallow data. Segregate­representation loss enlarges interclass distance by denying the unlabeled identities with low cosine similarity with labeled
identities. SimInstance helps the class center being more accurate by generating simulated instance feature. Then, it benefits the angular­margin based losses to guide the
model learned more discriminative feature representation. Our proposed SSRFace is easy
to apply and does not require additional training or parameters. Extensive experiments
have been conducted to prove our proposed method’s effectiveness.

[1] Y. Guo, L. Zhang, Y. Hu, X. He, and J. Gao, “Ms­celeb­1m: A dataset and benchmark
for large­scale face recognition,” CoRR, vol. abs/1607.08221, 2016.
[2] D. Yi, Z. Lei, S. Liao, and S. Z. Li, “Learning face representation from scratch,”
CoRR, vol. abs/1411.7923, 2014.
[3] Q. Cao, L. Shen, W. Xie, O. M. Parkhi, and A. Zisserman, “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, 2018.
[4] W. Liu, Y. Wen, Z. Yu, M. Li, B. Raj, and L. Song, “Sphereface: Deep hypersphere
embedding for face recognition,” in 2017 IEEE Conference on Computer Vision and
Pattern Recognition (CVPR), pp. 6738–6746, 2017.
[5] F. Wang, W. Liu, H. Liu, and J. Cheng, “Additive margin softmax for face verification,” CoRR, vol. abs/1801.05599, 2018.
[6] H. Wang, Y. Wang, Z. Zhou, X. Ji, D. Gong, J. Zhou, Z. Li, and W. Liu, “Cosface:
Large margin cosine loss for deep face recognition,” in Proceedings of the IEEE
conference on computer vision and pattern recognition, pp. 5265–5274, 2018.
[7] J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “Arcface: Additive angular margin loss for
deep face recognition,” in Proceedings of the IEEE/CVF Conference on Computer
Vision and Pattern Recognition, pp. 4690–4699, 2019.
[8] Z. Cheng, X. Zhu, and S. Gong, “Low­resolution face recognition,” in Asian Conference on Computer Vision, pp. 605–621, Springer, 2018.
[9] Z. Cheng, X. Zhu, and S. Gong, “Surveillance face recognition challenge,” arXiv
preprint arXiv:1804.09691, 2018.
[10] H. Du, H. Shi, Y. Liu, J. Wang, Z. Lei, D. Zeng, and T. Mei, “Semi­siamese training
for shallow face learning,” in European Conference on Computer Vision, pp. 36–53,
Springer, 2020.
[11] L. Yang, D. Chen, X. Zhan, R. Zhao, C. C. Loy, and D. Lin, “Learning to cluster
faces via confidence and connectivity estimation,” in Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern Recognition, pp. 13369–13378, 2020.
[12] L. Yang, X. Zhan, D. Chen, J. Yan, C. C. Loy, and D. Lin, “Learning to cluster faces
on an affinity graph,” in Proceedings of the IEEE/CVF Conference on Computer
Vision and Pattern Recognition, pp. 2298–2306, 2019.
[13] H. Yu, Y. Fan, K. Chen, H. Yan, X. Lu, J. Liu, and D. Xie, “Unknown identity
rejection loss: Utilizing unlabeled data for face recognition,” in Proceedings of the
IEEE/CVF International Conference on Computer Vision Workshops, pp. 0–0, 2019.
[14] X. Zhan, Z. Liu, J. Yan, D. Lin, and C. C. Loy, “Consensus­driven propagation in
massive unlabeled data for face recognition,” in Proceedings of the European Conference on Computer Vision (ECCV), pp. 568–583, 2018.
[15] W. Li, T. Guo, P. Li, B. Chen, B. Wang, W. Zuo, and L. Zhang, “Virface: Enhancing face recognition via unlabeled shallow data,” in Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern Recognition, pp. 14729–14738, 2021.
[16] T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in International conference on machine
learning, pp. 1597–1607, PMLR, 2020.
[17] Y. Feng, F. Wu, X. Shao, Y. Wang, and X. Zhou, “Joint 3d face reconstruction and
dense alignment with position map regression network,” in Proceedings of the European Conference on Computer Vision (ECCV), pp. 534–551, 2018.
[18] D. S. Tan, J. H. Soeseno, and K.­L. Hua, “Controllable and identity­aware facial
attribute transformation,” IEEE Transactions on Cybernetics, pp. 1–12, 2021.
[19] Y. Deng, J. Yang, S. Xu, D. Chen, Y. Jia, and X. Tong, “Accurate 3d face reconstruction with weakly­supervised learning: From single image to image set,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
Workshops, pp. 0–0, 2019.
[20] Y. Liu, F. Wei, J. Shao, L. Sheng, J. Yan, and X. Wang, “Exploring disentangled
feature representation beyond face identification,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2080–2089, 2018.
[21] K. He, H. Fan, Y. Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the IEEE/CVF Conference
on Computer Vision and Pattern Recognition, pp. 9729–9738, 2020.
[22] Y. Wu, H. Liu, and Y. Fu, “Low­shot face recognition with hybrid classifiers,” in
Proceedings of the IEEE International Conference on Computer Vision Workshops,
pp. 1933–1939, 2017.
[23] L. Wang, Y. Li, and S. Wang, “Feature learning for one­shot face recognition,” in
2018 25th IEEE International Conference on Image Processing (ICIP), pp. 2386–
2390, IEEE, 2018.
[24] Z. Ding, Y. Guo, L. Zhang, and Y. Fu, “Generative one­shot face recognition,” arXiv
preprint arXiv:1910.04860, 2019.
[25] Y. Guo and L. Zhang, “One­shot face recognition by promoting underrepresented
classes,” arXiv preprint arXiv:1707.05574, 2017.
[26] C.­H. Hsu, K.­L. Hua, and W.­H. Cheng, “Physiognomy master: a novel personality
analysis system based on facial features,” in Proceedings of the 21st ACM international conference on Multimedia, pp. 407–408, 2013.
[27] P. Chandran, D. Bradley, M. Gross, and T. Beeler, “Attention­driven cropping for
very high resolution facial landmark detection,” in Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern Recognition (CVPR), June 2020.
[28] Z. Liu, X. Zhu, G. Hu, H. Guo, M. Tang, Z. Lei, N. M. Robertson, and J. Wang,
“Semantic alignment: Finding semantically consistent ground­truth for facial landmark detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision
and Pattern Recognition (CVPR), June 2019.
[29] F. Schroff, D. Kalenichenko, and J. Philbin, “Facenet: A unified embedding for face
recognition and clustering,” in Proceedings of the IEEE conference on computer
vision and pattern recognition, pp. 815–823, 2015.
[30] M.­C. Hu, H.­T. Wu, L.­Y. Lo, T.­Y. Pan, W.­H. Cheng, K.­L. Hua, and T. Mei, “A
framework of enlarging face datasets used for makeup face analysis,” in 2016 IEEE
Second International Conference on Multimedia Big Data (BigMM), pp. 219–222,
2016.
[31] K. Sohn, “Improved deep metric learning with multi­class n­pair loss objective,” in
Advances in neural information processing systems, pp. 1857–1865, 2016.
[32] Y. Sun, Deep learning face representation by joint identification­verification. The
Chinese University of Hong Kong (Hong Kong), 2015.
[33] G. B. Huang, M. Mattar, T. Berg, and E. Learned­Miller, “Labeled faces in the wild:
A database forstudying face recognition in unconstrained environments,” in Workshop on faces in’Real­Life’Images: detection, alignment, and recognition, 2008.
[34] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,”
in Proceedings of the IEEE conference on computer vision and pattern recognition,
pp. 770–778, 2016.
[35] F. V. Massoli, G. Amato, and F. Falchi, “Cross­resolution learning for face recognition,” Image and Vision Computing, vol. 99, p. 103927, 2020.
[36] Y. Shi, X. Yu, K. Sohn, M. Chandraker, and A. K. Jain, “Towards universal representation learning for deep face recognition,” in Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern Recognition, pp. 6817–6826, 2020.
[37] S. S. Khalid, M. Awais, Z.­H. Feng, C.­H. Chan, A. Farooq, A. Akbari, and J. Kittler,
“Resolution invariant face recognition using a distillation approach,” IEEE Transactions on Biometrics, Behavior, and Identity Science, vol. 2, no. 4, pp. 410–420,
2020.
[38] T. Zheng, W. Deng, and J. Hu, “Cross­age lfw: A database for studying cross­age
face recognition in unconstrained environments,” arXiv preprint arXiv:1708.08197,
2017.
[39] T. Zheng and W. Deng, “Cross­pose lfw: A database for studying cross­pose
face recognition in unconstrained environments,” Beijing University of Posts and
Telecommunications, Tech. Rep, vol. 5, p. 7, 2018.
[40] S. Sengupta, J.­C. Chen, C. Castillo, V. M. Patel, R. Chellappa, and D. W. Jacobs,
“Frontal to profile face verification in the wild,” in 2016 IEEE Winter Conference
on Applications of Computer Vision (WACV), pp. 1–9, IEEE, 2016.