臺灣博碩士論文加值系統

　　近年來人臉辨識科技已經非常成熟，並廣泛的應用在日常生活中，如：電子 支付、智能住宅、手機螢幕解鎖……等。而與科技便利性伴隨而來的是資訊安全 問題，人臉資訊可能遭到偽造、篡改…等惡意的電子欺騙攻擊(Spoofing Attack)，造成使用者的疑慮，有心人士可能只要取得一張使用者的照片便可以偽裝成使用 者，並竊取使用者的隱私資訊。為了對抗這種惡意攻擊，臉部防偽(face anti-spoofing )便成為值得重視的議題。本研究提出一種即時的人臉防偽及辨識系統，分為兩系統，採用 Mask R-CNN(Mask Region-based Convolutional Neural Networks)對防偽系統的訓練，透過輸入標記資料使防偽系統可以區分出人臉及手機的特徵，以避免將手機中的相片誤判為人臉的情況，再來，針對防偽的結果 進行人臉辨識，辨識系統透過與資料庫中的使用者人臉特徵向量比對計算歐式距離以判斷否為合法使用者，目標是在不使用特殊硬體設備及額外的驗證系統，也可以做到防偽及辨識的功能。

　　In recent years, face recognition technology has been very mature and widely used in our daily life, such as: electronic payment, smart home, screen unlocking for mobile phone, etc.
　　The emerging technologies can provide more convenience lift to user, but it always ignores the issue of information security. The user’s face information might be subjected to some spoofing attacks, such as: forgery, tampering, and so on, leading to losing the confidence of users. Adversary can take a photo with user face information from the anywhere, it can camouflage as a legal user to steal user private information. For avoiding the illegal access, face anti-spoofing became more and more import topic.
　　In this thesis, the system of anti-spoofing is proposed. The Mask R-CNN approach is used to train the weight dataset in the proposed system. The proposed system is performed, and then, the face recognition result is obtained according to training dataset and predicted result. In particular, the goal of this study is to achieve the higher face recognition rate in the system of anti-spoofing and identification without using specified hardware equipment.

誌謝　I
摘要　II
Abstract　III
目錄　IV
圖目錄　VI
表目錄　VIII

1.緒論　1
　1.1 研究背景及動機　1
　1.2 研究目的　2
2.相關研究　4
　2.1 監督式學習　4
　2.2 Mask R-CNN的發展　5
　　2.2.1 R-CNN　6
　　2.2.2 Fast R-CNN　7
　　2.2.3 Faster R-CNN　10
　　2.2.4 Mask R-CNN　14
　2.3 Dlib函式庫介紹　19
　　2.3.1 Dlib 人臉辨識模型　21
　2.4 人臉追蹤、辨識、防偽之相關研究探討　21
3.研究方法　22
　3.1 動態人臉偵測　23
　3.2 開運算(opening)　23
　3.3 Mask R-CNN模型預測　24
　　3.3.1 訓練Mask R-CNN模型　24
　　3.3.2 監督式訓練　24
　3.4 訓練結果預測　27
　3.5 建立使用者人臉比對資料庫　30
　3.6 人臉辨識系統　31
4.研究結果與分析　32
　4.1 實驗限制及方式　32
　4.2 訓練資料集與前置作業　32
　4.3 模型評估標準　32
　4.4 模型預測結果評估　33
　4.5 人臉防偽實驗　34
　　4.5.1 相片攻擊　34
　　4.5.2 人臉辨識　36
5.結論與建議　44
參考文獻　45

外文文獻
[1]J. Li , Y. Wang , T. Tan , A .K . Jain , “Live Face Detection Based on the Analysis of Fourier Spectra,” Proceedings of the International Society for Optical Engineering (SPIE), Vol. 5404, pp. 296-303, 2004.
[2]K. He, G. Gkioxari, P. Dollár and R. Girshick, “Mask R-CNN,” Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV), pp. 2980-2988, 2017.
[3]Dlib C++ Library source, http://dlib.net/ Accessed by 18 March 2020.
[4]Cortes, C., Vapnik, V. “Support-vector networks,” Machine Learning, Vol. 20, pp. 273-297,1995.
[5]Wada, K., labelme: Image Polygonal Annotation with Python. 2016.
[6]Russell, B.C., Torralba, A., Murphy, K.P. et al. “LabelMe: A Database and Web-Based Tool for Image Annotation,” International Journal of Computer Vision ,” Vol. 77, pp. 157-173, 2008.
[7]Unsplash, https://unsplash.com/ Accessed by 18 March 2020.
[8]R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014.
[9]J. R. R. Uijlings , K. E. A. van de Sande , T. Gevers , A. W. M. Smeulders, “Selective Search for Object Recognition,” International Journal of Computer Vision, Vol.104, pp. 154-171, 2013.
[10]A. Krizhevsky, I. Sutskever, and G. Hinton. “ImageNet classification with deep convolutional neural networks,” Proceedings of the 25th International Conference on Neural Information Processing Systems, pp. 1097-1105, 2012.
[11]R. Girshick, “Fast R-CNN,” Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV), pp. 1440-1448, 2015.
[12]K. Simonyan and A. Zisserman. “Very deep convolutional networks for large-scale image recognition,” Proceedings of the International Conference on Learning Representations (ICLR),pp. 1-14, 2015.
[13]S. Ren, K. He, R. Girshick and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 39, pp. 1137-1149, 2017.
[14]K. He , X. Zhang, S. Ren and J. Sun, “Deep Residual Learning for Image Recognition,” Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.770-778, 2016.
[15]T. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, “Feature Pyramid Networks for Object Detection,” Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 936-944, 2017.
[16]M. Jaderberg, K. Simonyan, A. Zisserman et al., “Spatial Transformer Networks,” Proceedings of the Advances in Neural Information Processing Systems (NIPS), 2015.
[17]dlib source, http://dlib.net/face_recognition.py.html. Accessed by 18 March 2020.
[18]Mathematicalmorphology, https://en.wikipedia.org/wiki/Mathematical_morphology. Accessed by 18 March 2020.
[19]COCO evaluate, http://cocodataset.org/#detection-eval,Accessed by 8 April 2020.

中文文獻
[20]江思賢，2018，「基於熱影像特徵之多人臉辨識系統」，國立臺北科技大學，碩
士論文。
[21]李亭緯，2008，「利用人臉五官為特徵之人臉辨識系統」，國立中央大學，碩士論文。
[22]林仁信，2018，「基於深度學習人臉辨識理論於桌上型裝置系統」，國立暨南國際大學，碩士論文。