人臉辨識的應用隨著深度學習演算法的發展與硬體晶片加速神經網路的計算而越來越普及。在人臉辨識中，辨識準確率容易受到光線、距離與遮蔽的影響。其中遮蔽的影響是最難克服的，遮蔽的影響會使得辨識結果錯誤，甚至連臉部偵測都偵測不到。因此目前人臉辨識系統都侷限於限制環境的範圍下操作。本篇論文使用卷積神經網路訓練強健的人臉偵測網路以及臉部特徵擷取網路，提升人臉偵測的準確率與臉部特徵擷取的能力。克服了人臉在遭受遮蔽的情況下無法正常辨識的情況。
本篇論文使用網路攝影機擷取影像作為系統之輸入，經由人臉偵測網路計算出影像中所有人臉區域以及臉部標記點。透過臉部標記點進行臉部校正後再輸入人臉辨識網路判定身分。人臉偵測的網路使用特徵金字塔減少參數並且達到尺度不變性。搭配不同檢測模塊針對不同尺度的人臉進行偵測，並且結合影像脈絡模塊增加感受野，提升偵測準確率並減少記憶體使用。臉部特徵擷取網路在訓練階段導入附加角度邊界損失函數，可以有效擴大不同種類之間分類的邊界距離，透過這種損失函數學習機制能有效且快速的訓練資料庫人物臉部特徵，提升分類器準確率。
本篇論文使用WIDER資料庫、MS-Celeb-1M的資料庫訓練，在測試部分使用自己蒐集的資料庫。實驗結果顯示在人臉無遮蔽的情況下可以達到97.36%的準確率。人臉遮蔽25%、50%的情況下達到了96.15%、88.46%的準確率。本系統確實提升了人臉辨識準確率並且克服人臉在遭受遮蔽的情況下無法正常辨識的問題。

The application of face recognition has become more and more popular with the development of deep learning algorithms and the calculation of hardware chips for accelerating neural networks. In face recognition, the recognition accuracy is easily affected by light, distance and occlusion. However, the effect of the occlusion is the most difficult to overcome. The effect of the occlusion will make the recognition result wrong, and even the face will not be detected. Therefore, the current face recognition system is limited to operate under the constraint environment. This thesis uses a convolutional neural network to train a robust face detection network and facial feature extraction network to improve the accuracy of face detection and the ability to capture facial features. Therefore, the method overcomes the situation that the face cannot be recognized normally under the occlusion.
The system uses a webcam to capture images as input to the system. All face regions and facial landmark in the image are calculated via the face detection network. The face is aligned by facial landmark and then input into the face recognition network to determine the identity. Face detection network uses feature pyramids to reduce parameters and achieve scale invariance. Different detection module are used to detect different scale faces. Besides,we combine the image context module to increase the receptive field. It improves detection
accuracy and reduce memory usage effectively.
This paper uses the WIDER database, MS-Celeb-1M database for training, and uses the database that I collected in the test section. The experimental result of the thesis show that 97.36% accuracy can be reached without occlusion. The accuracy of 96.15% and 88.46% is achieved when the face is under 25% and 50% occlusion. The proposed system does improve the face recognition accuracy and overcomes the problem that the face cannot be recognized normally when it is occluded.

[1] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks." pp. 1097-1105.
[2] N. Dalal, and B. Triggs, "Histograms of oriented gradients for human detection."
[3] D. G. Lowe, “Distinctive image features from scale-invariant keypoints,” International journal of computer vision, vol. 60, no. 2, pp. 91-110, 2004.
[4] J. P. Jones, and L. A. Palmer, “An evaluation of the two-dimensional Gabor filter model of simple receptive fields in cat striate cortex,” Journal of neurophysiology, vol. 58, no. 6, pp. 1233-1258, 1987.
[5] T. Ojala, M. Pietikäinen, and T. Mäenpää, “Multiresolution gray-scale and rotation invariant texture classification with local binary patterns,” IEEE Transactions on Pattern Analysis & Machine Intelligence, no. 7, pp. 971-987, 2002.
[6] G. Yang, and T. S. Huang, “Human face detection in a complex background,” Pattern recognition, vol. 27, no. 1, pp. 53-63, 1994.
[7] T. A. McGregor, R. L. Klatzky, C. Hamilton, and S. J. Lederman, “Haptic classification of facial identity in 2D displays: Configural versus feature-based processing,” IEEE transactions on haptics, vol. 3, no. 1, pp. 48-55, 2009.
[8] C.-T. Tu, and J.-J. J. Lien, “Automatic location of facial feature points and synthesis of facial sketches using direct combined model,” IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 40, no. 4, pp. 1158-1169, 2009.
[9] K. Sandeep, and A. Rajagopalan, "Human Face Detection in Cluttered Color Images Using Skin Color, Edge Information."
[10] R. Brunelli, and T. Poggio, “Face recognition: Features versus templates,” IEEE transactions on pattern analysis and machine intelligence, vol. 15, no. 10, pp. 1042-1052, 1993.
[11] A. K. Jain, Y. Zhong, and M.-P. Dubuisson-Jolly, “Deformable template models: A review,” Signal processing, vol. 71, no. 2, pp. 109-129, 1998.
[12] B. Moghaddam, and A. P. Pentland, "Face recognition using view-based and modular eigenspaces." pp. 12-21.
[13] Y. Fu, and N. Zheng, “M-face: An appearance-based photorealistic model for multiple facial attributes rendering,” IEEE Transactions on Circuits and Systems for Video technology, vol. 16, no. 7, pp. 830-842, 2006.
[14] F. Dornaika, and F. Davoine, “On appearance based face and facial action tracking,” IEEE transactions on circuits and systems for video technology, vol. 16, no. 9, pp. 1107-1124, 2006.
[15] K. K. Sung, and T. Poggio, Example Based Learning for View-Based Human Face Detection, MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB, 1994.
[16] M. Kirby, and L. Sirovich, “Application of the Karhunen-Loeve procedure for the characterization of human faces,” IEEE Transactions on Pattern analysis and Machine intelligence, vol. 12, no. 1, pp. 103-108, 1990.
[17] J. Miao, B. Yin, K. Wang, L. Shen, and X. Chen, “A hierarchical multiscale and multiangle system for human face detection in a complex background using gravity-center template,” Pattern Recognition, vol. 32, no. 7, pp. 1237-1248, 1999.
[18] F. Rosenblatt, “The perceptron: a probabilistic model for information storage and organization in the brain,” Psychological review, vol. 65, no. 6, pp. 386, 1958.
[19] Y. LeCun, B. Boser, J. S. Denker, D. Henderson, R. E. Howard, W. Hubbard, and L. D. Jackel, “Backpropagation applied to handwritten zip code recognition,” Neural computation, vol. 1, no. 4, pp. 541-551, 1989.
[20] M. A. Hearst, S. T. Dumais, E. Osuna, J. Platt, and B. Scholkopf, “Support vector machines,” IEEE Intelligent Systems and their applications, vol. 13, no. 4, pp. 18-28, 1998.
[21] K. Simonyan, and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556, 2014.
[22] C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, "Going deeper with convolutions." pp. 1-9.
[23] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition." pp. 770-778.
[24] M. Tan, and Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” arXiv preprint arXiv:1905.11946, 2019.
[25] C. Garcia, and M. Delakis, “Convolutional face finder: A neural architecture for fast and robust face detection,” IEEE Transactions on pattern analysis and machine intelligence, vol. 26, no. 11, pp. 1408-1423, 2004.
[26] M. Osadchy, Y. L. Cun, and M. L. Miller, “Synergistic face detection and pose estimation with energy-based models,” Journal of Machine Learning Research, vol. 8, no. May, pp. 1197-1215, 2007.
[27] D. Chen, S. Ren, Y. Wei, X. Cao, and J. Sun, "Joint cascade face detection and alignment." pp. 109-122.
[28] C. Zhang, and Z. Zhang, "Improving multiview face detection with multi-task deep convolutional neural networks." pp. 1036-1041.
[29] C. Zhu, Y. Zheng, K. Luu, and M. Savvides, "Cms-rcnn: contextual multi-scale region-based cnn for unconstrained face detection," Deep Learning for Biometrics, pp. 57-79: Springer, 2017.
[30] 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.
[31] G. B. Huang, M. Mattar, T. Berg, and E. Learned-Miller, "Labeled faces in the wild: A database forstudying face recognition in unconstrained environments."
[32] N. Kumar, A. C. Berg, P. N. Belhumeur, and S. K. Nayar, "Attribute and simile classifiers for face verification." pp. 365-372.
[33] Y. Taigman, M. Yang, M. A. Ranzato, and L. Wolf, "Deepface: Closing the gap to human-level performance in face verification." pp. 1701-1708.
[34] F. Schroff, D. Kalenichenko, and J. Philbin, "Facenet: A unified embedding for face recognition and clustering." pp. 815-823.
[35] J. Hu, J. Lu, and Y.-P. Tan, "Robust partial face recognition using instance-to-class distance." pp. 1-6.
[36] S. Liao, A. K. Jain, and S. Z. Li, “Partial face recognition: Alignment-free approach,” IEEE Transactions on pattern analysis and machine intelligence, vol. 35, no. 5, pp. 1193-1205, 2012.
[37] J. Wright, A. Y. Yang, A. Ganesh, S. S. Sastry, and Y. Ma, “Robust face recognition via sparse representation,” IEEE transactions on pattern analysis and machine intelligence, vol. 31, no. 2, pp. 210-227, 2008.
[38] Y. Su, Y. Yang, Z. Guo, and W. Yang, "Face recognition with occlusion." pp. 670-674.
[39] R. Weng, J. Lu, and Y.-P. Tan, “Robust point set matching for partial face recognition,” IEEE transactions on image processing, vol. 25, no. 3, pp. 1163-1176, 2016.
[40] L. He, H. Li, Q. Zhang, Z. Sun, and Z. He, "Multiscale representation for partial face recognition under near infrared illumination." pp. 1-7.
[41] L. He, H. Li, Q. Zhang, and Z. Sun, "Dynamic feature learning for partial face recognition." pp. 7054-7063.
[42] M. Najibi, P. Samangouei, R. Chellappa, and L. S. Davis, "Ssh: Single stage headless face detector." pp. 4875-4884.
[43] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y. Fu, and A. C. Berg, "Ssd: Single shot multibox detector." pp. 21-37.
[44] R. Girshick, J. Donahue, T. Darrell, and J. Malik, "Rich feature hierarchies for accurate object detection and semantic segmentation." pp. 580-587.
[45] R. Girshick, "Fast r-cnn." pp. 1440-1448.
[46] S. Ren, K. He, R. Girshick, and J. Sun, "Faster r-cnn: Towards real-time object detection with region proposal networks." pp. 91-99.
[47] J. Deng, J. Guo, N. Xue, and S. Zafeiriou, "Arcface: Additive angular margin loss for deep face recognition." pp. 4690-4699.
[48] O. M. Parkhi, A. Vedaldi, and A. Zisserman, "Deep face recognition." p. 6.
[49] W. Liu, Y. Wen, Z. Yu, M. Li, B. Raj, and L. Song, "Sphereface: Deep hypersphere embedding for face recognition." pp. 212-220.
[50] R. Ranjan, C. D. Castillo, and R. Chellappa, “L2-constrained softmax loss for discriminative face verification,” arXiv preprint arXiv:1703.09507, 2017.
[51] 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.
[52] Y. Liu, H. Li, and X. Wang, “Rethinking feature discrimination and polymerization for large-scale recognition,” arXiv preprint arXiv:1710.00870, 2017.
[53] F. Wang, X. Xiang, J. Cheng, and A. L. Yuille, "Normface: l 2 hypersphere embedding for face verification." pp. 1041-1049.
[54] 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." pp. 5265-5274.