跳到主要內容

臺灣博碩士論文加值系統

(44.192.67.10) 您好!臺灣時間:2024/11/09 17:50
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:王乃民
研究生(外文):Wang, Nai-Min
論文名稱:人臉去識別化與生成對抗式神經網絡
論文名稱(外文):Face De-identification with GAN
指導教授:高竹嵐高竹嵐引用關係
指導教授(外文):Kao, Chu-Lan
口試委員:鄭又仁簡仁宗王釧茹高竹嵐
口試委員(外文):Cheng, Yu-JenChien, Jen-TzungWang, Chuan-JuKao, Chu-Lan
口試日期:2019-06-24
學位類別:碩士
校院名稱:國立交通大學
系所名稱:統計學研究所
學門:數學及統計學門
學類:統計學類
論文種類:學術論文
論文出版年:2019
畢業學年度:107
語文別:英文
論文頁數:29
中文關鍵詞:去識別化生成對抗神經網絡影像處理
外文關鍵詞:face de-identificationGANimage process
相關次數:
  • 被引用被引用:0
  • 點閱點閱:219
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
近幾年來, 影像發展蓬勃, 不論是娛樂、生活或是教育, 影像佔
了很大的比例, 然而在某些場合人們需要保留隱私, 現今常見的處
理是馬賽克, 而馬賽克會影響觀看的視覺觀感, 加上卷積神經網絡
的進步, 已經進步到可透過除去馬賽克還原圖片, 人們的隱私將會
受到危害, 因此希望透過這篇論文來討論透過生成對抗式神經網絡
的模型來加強人臉去識別化的效果, 主要是透過設計網絡架構以及
函數的變化來達到去識別化的目的, 提出二次投影的方法希望將原
始的照片映射到去識別化空間。
Recently, images and videos are explosively growing no matter in
entertainment, education or daily life. They take lots of proportion in
our lives. However, people do not like to be recognized sometimes. They
need to keep their identities in privacy. The common method to deal
with this situation is pixelation which may make the audience feel bad.
Besides, the convolution network improves to restore pixilated parts
and recognize the original identification, which is a threat for people
who want to keep their privacy. Therefore, this paper targets to discuss
de-identification with GAN through designing the architecture of the
network and objective function with better performance. In particular,
we propose a twice mapping method, which essentially maps the original
image to a de-identified fact. As far as know, this is the first time that
such kind of method has been proposed in the de-identification
literature.
1. Introduction 1
1.1. Background 1
1.2. Related Work 2
2. Implement 3
2.1. GAN 3
2.2. Dataset 5
2.3. Problem 6
2.4. Objective and Structure 8
3. Experiment 13
3.1. Pixelization case 13
3.2. Penalty case 14
3.3. Mapping case 19
4. Comparison of methods 21
5. Conclusions and Discussions 23
6. Reference 24
7. Appendix 26
[1] Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville A. & Bengio, Y. (2014). Generative adversarial nets. In Advances in neural information processing systems (pp. 2672-2680).
[2] Wei, X., Gong, B., Liu, Z., Lu, W., & Wang, L. (2018). Improving the improved training of Wasserstein Gans: A consistency term and its dual effect. arXiv preprint arXiv:1803.01541.
[3] Mirza, M., & Osindero, S. (2014). Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784.
[4] Choi, Y., Choi, M., Kim, M., Ha, J. W., Kim, S., & Choo, J. (2018). StarGan: Unified generative adversarial networks for multi-domain image-to-image translation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 8789-8797).
[5] Samarzija, B., & Ribaric, S. (2014, May). An approach to the de-identification of faces in different poses. In 2014 37th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO) (pp. 1246-1251). IEEE.
[6] Wu, Y., Yang, F., & Ling, H. (2018). Privacy-Protective-GAN for Face De-identification. arXiv preprint arXiv:1806.08906.
[7] Dosovitskiy, A., Tobias Springenberg, J., & Brox, T. (2015). Learning to generate chairs with convolutional neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 1538-1546).
[8] Arjovsky, M., Chintala, S., & Bottou, L. (2017). Wasserstein gan. arXiv preprint arXiv:1701.07875.
[9] Kellerer, H. G. (1985). Duality theorems and probability metrics. In Proceedings of the seventh conference on probability theory (Brasov, 1982) (pp. 211-220).
[10] Schroff, F., Kalenichenko, D., & Philbin, J. (2015). Facenet: A unified embedding for face recognition and clustering. In Proceedings of the IEEE conference on computer vision and pattern recognition
[11] LeFevre, K., DeWitt, D. J., & Ramakrishnan, R. (2006, April). Mondrian multidimensional k-anonymity. In ICDE (Vol. 6, p. 25).
[12] Cootes, T. F., & Taylor, C. J. (2004). Statistical models of appearance for computer vision.
[13] Meden, B., Mallı, R. C., Fabijan, S., Ekenel, H. K., Štruc, V., & Peer, P. (2017). Face deidentification with generative deep neural networks. IET Signal Processing, 11(9), 1046-1054.
[14] Noh, H., Hong, S., & Han, B. (2015). Learning deconvolution network for semantic segmentation. In Proceedings of the IEEE international conference on computer vision (pp. 1520-1528).
[15] Isola, P., Zhu, J. Y., Zhou, T., & Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1125-1134).
[16] Odena, A. (2016). Semi-supervised learning with generative adversarial networks. arXiv preprint arXiv:1606.01583.
[17] P Wei, L., Zhang, S., Gao, W., & Tian, Q. (2018). Person transfer gan to bridge domain gap for person re-identification. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 79-88).
[18] Radford, A., Metz, L., & Chintala, S. (2015). Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434.
[19] Karras, T., Laine, S., & Aila, T. (2018). A style-based generator architecture for generative adversarial networks. arXiv preprint arXiv:1812.04948.
連結至畢業學校之論文網頁點我開啟連結
註: 此連結為研究生畢業學校所提供,不一定有電子全文可供下載,若連結有誤,請點選上方之〝勘誤回報〞功能,我們會盡快修正,謝謝!
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top