過去的幾年發布了許多人臉辨識模型，隨之而來的還有許多文獻和開源的模型供大家利用。雖然已經有許多人臉辨識模型可以使用，不過我們發現仍然有一個問題 — 尚沒有比較全面的基於實驗數據來做比較的人臉辨識的審閱工作。這讓人臉辨識領域的學者們難以針對不同的應用場域採用適當的人臉辨識模型模型。出於上述原因，我們選用了7個不同的開源人臉辨識的方法加以評估。針對每一個方法，我們選用一個當前效能最好的模型；針對每一個模型，我們進行相同標準的人臉驗證 (face verification)和人臉認證 (face identification) 測試實驗。人臉驗證實驗是一對一地比較測試，目的在於直接測試模型的分辨 (discriminative)能力；而人臉認真實驗是一對多的模擬搜尋實驗，目的在於模擬在真實情境中，模型從一個大集合 (gallery)中找尋所需所給定目標 (probe)的任務。我們採用IJB-C數據集作為人臉驗證實驗的測試集，Megaface數據集作為人臉認證實驗的測試集。我們同時展現我們所做的測試結果和各個方法在原本文獻中的測試結果，以作為比較之用。我們的實驗主要基於測試軟體平台Insightface（與Arcface這個模型一併發佈）而進行的。我們根據自己的需要做了小幅度的修改。我們的研究可以看作上述軟體平台的延伸工作。我們希望我們的研究可以給未來的學者們提供寶貴的意見，並且可以被作為日後學者們的研究基準 (baseline)。

Numerous FR methods have been released in the past several years, as relating papers have widely published and plenty of them have open-source models free to use. Though there are several options for FR tasks, there lies problem – it lacks of profound evidence-based research on comparing these FR models, which put into a situation that researches may not easily find out the most suitable solution for typical applications. For the reason mentioned above, we conduct experimental evaluation on the 7 different open-source solutions. We choose the best performed model as the representative of each method, and for each elected model, we evaluate the true performances under the same standard, which consists of a face verification task and a face identification task. The face verification task is a pairwise one-to-one test, validating the discriminative power of a model. The face identification task, on the other hand, is an open set problem simulating the real scenario of suspect within a large pool of gallery by referencing a probe image. We perform the verification task and identification task using IJB-C [1] and Megaface [2] dataset, respective. The evaluation result of each model is presented, along with the statistics published by each one’s original paper. The majority of our work is powered by the Insightface testpack – a software released along with the most popular Arcface [3] online resource – with some modification to meet our requirement. Our work can be treated as the extended investigation using the aforementioned and we are sure the same workflow can be used for future studies. Our work presents an honest comparison of most recent deep learning-based solutions and has the potential to be referred as a baseline for the following researchers.

[1] B. Maze et al., "IARPA janus benchmark-c: Face dataset and protocol," in 2018 International Conference on Biometrics (ICB), 2018: IEEE, pp. 158-165.
[2] I. Kemelmacher-Shlizerman, S. M. Seitz, D. Miller, and E. Brossard, "The megaface benchmark: 1 million faces for recognition at scale," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 4873-4882.
[3] J. Deng, J. Guo, N. Xue, and S. Zafeiriou, "Arcface: Additive angular margin loss for deep face recognition," arXiv preprint arXiv:1801.07698, 2018.
[4] M. Wang and W. Deng, "Deep face recognition: A survey," arXiv preprint arXiv:1804.06655, 2018.
[5] C. Ding and D. Tao, "A comprehensive survey on pose-invariant face recognition," ACM Transactions on intelligent systems and technology (TIST), vol. 7, no. 3, p. 37, 2016.
[6] X. Jin and X. Tan, "Face alignment in-the-wild: A survey," Computer Vision and Image Understanding, vol. 162, pp. 1-22, 2017.
[7] Y. Wen, K. Zhang, Z. Li, and Y. Qiao, "A discriminative feature learning approach for deep face recognition," in European conference on computer vision, 2016: Springer, pp. 499-515.
[8] F. Wang, X. Xiang, J. Cheng, and A. L. Yuille, "Normface: l 2 hypersphere embedding for face verification," in Proceedings of the 25th ACM international conference on Multimedia, 2017: ACM, pp. 1041-1049.
[9] W. Liu, Y. Wen, Z. Yu, M. Li, B. Raj, and L. Song, "Sphereface: Deep hypersphere embedding for face recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 212-220.
[10] H. Wang et al., "Cosface: Large margin cosine loss for deep face recognition," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 5265-5274.
[11] S. Chen, Y. Liu, X. Gao, and Z. Han, "Mobilefacenets: Efficient cnns for accurate real-time face verification on mobile devices," in Chinese Conference on Biometric Recognition, 2018: Springer, pp. 428-438.
[12] K. Zhao, J. Xu, and M.-M. Cheng, "Regularface: Deep face recognition via exclusive regularization," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 1136-1144.
[13] Y. Duan, J. Lu, and J. Zhou, "Uniformface: Learning deep equidistributed representation for face recognition," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 3415-3424.
[14] X. Zhang, R. Zhao, Y. Qiao, X. Wang, and H. Li, "Adacos: Adaptively scaling cosine logits for effectively learning deep face representations," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 10823-10832.
[15] Y. Taigman, M. Yang, M. A. Ranzato, and L. Wolf, "Deepface: Closing the gap to human-level performance in face verification," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 1701-1708.
[16] 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, 2015, pp. 815-823.
[17] C. Liu and H. Wechsler, "Gabor feature based classification using the enhanced fisher linear discriminant model for face recognition," IEEE Transactions on Image processing, vol. 11, no. 4, pp. 467-476, 2002.
[18] W. Zhang, S. Shan, W. Gao, X. Chen, and H. Zhang, "Local Gabor binary pattern histogram sequence (LGBPHS): a novel non-statistical model for face representation and recognition," in Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1, 2005, vol. 1: IEEE, pp. 786-791.
[19] L. Wolf, T. Hassner, and I. Maoz, Face recognition in unconstrained videos with matched background similarity. IEEE, 2011.
[20] M. Kim, S. Kumar, V. Pavlovic, and H. Rowley, "Face tracking and recognition with visual constraints in real-world videos," in 2008 IEEE Conference on Computer Vision and Pattern Recognition, 2008: IEEE, pp. 1-8.
[21] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
[22] O. Russakovsky et al., "Imagenet large scale visual recognition challenge," International journal of computer vision, vol. 115, no. 3, pp. 211-252, 2015.
[23] G. B. Huang, M. Mattar, T. Berg, and E. Learned-Miller, "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.
[24] Y. Sun, X. Wang, and X. Tang, "Deep learning face representation from predicting 10,000 classes," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 1891-1898.
[25] O. M. Parkhi, A. Vedaldi, and A. Zisserman, "Deep face recognition," in bmvc, 2015, vol. 1, no. 3, p. 6.
[26] Y. Sun, D. Liang, X. Wang, and X. Tang, "Deepid3: Face recognition with very deep neural networks," arXiv preprint arXiv:1502.00873, 2015.
[27] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[28] 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, 2016, pp. 770-778.
[29] X. Wu, R. He, Z. Sun, and T. Tan, "A light cnn for deep face representation with noisy labels," IEEE Transactions on Information Forensics and Security, vol. 13, no. 11, pp. 2884-2896, 2018.
[30] C. Szegedy et al., "Going deeper with convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1-9.
[31] F. Wang, J. Cheng, W. Liu, and H. Liu, "Additive margin softmax for face verification," IEEE Signal Processing Letters, vol. 25, no. 7, pp. 926-930, 2018.
[32] D. Yi, Z. Lei, S. Liao, and S. Z. Li, "Learning face representation from scratch," arXiv preprint arXiv:1411.7923, 2014.
[33] 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), 2018: IEEE, pp. 67-74.
[34] Y. Guo, L. Zhang, Y. Hu, X. He, and J. Gao, "Ms-celeb-1m: A dataset and benchmark for large-scale face recognition," in European Conference on Computer Vision, 2016: Springer, pp. 87-102.
[35] K. Zhang, Z. Zhang, Z. Li, and Y. Qiao, "Joint face detection and alignment using multitask cascaded convolutional networks," IEEE Signal Processing Letters, vol. 23, no. 10, pp. 1499-1503, 2016.
[36] B. Thomee et al., "The new data and new challenges in multimedia research," arXiv preprint arXiv:1503.01817, vol. 1, no. 8, 2015.
[37] T. Chen et al., "Mxnet: A flexible and efficient machine learning library for heterogeneous distributed systems," arXiv preprint arXiv:1512.01274, 2015.
[38] Y. Jia et al., "Caffe: Convolutional architecture for fast feature embedding," in Proceedings of the 22nd ACM international conference on Multimedia, 2014, pp. 675-678.