這篇論文介紹利用深度學習，學習地圖藝術的風格，無論是在使用者給的新地圖上或是原本地圖藝術的背景地圖，我們都能夠將肖像貼到地圖上，自動學習並產生新的地圖藝術作品。我們這篇的方法有兩個部分：1. 取出地圖藝術中的肖像，改良風格轉換的方法，將輸入的照片與取出的肖像做風格轉換。2. 透過深度卷積神經網路，並且利用增加損失函數和限制，將肖像貼上地圖，產生地圖藝術。而透過相同的方法，我們先對每張圖的值做分類，找出肖像位置，再與圖像資料一起做訓練，將地圖藝術中的肖像去除，恢復背景地圖。讓地圖藝術不僅能在新地圖上產生，也能在原本的地圖藝術上做不同肖像的變化。由於我們無法獲得正確的地圖藝術中的背景地圖，也沒有大量與地圖藝術上肖像相同風格的肖像圖像資料，因此在訓練過程中，訓練資料是找尋類似風格的地圖和肖像，將它們合成拿去訓練，透過此法逼近出結果。另外，由於產生出來的地圖結果還尚未為最佳，我們透過改良風格轉換的方法，將恢復的地圖做優化，達到最好的成果。

This paper introduces using deep learning to learn the style of map art. We can attach the portrait to the map given by the user or the background map of the original map art. Our method has two parts: 1. Get the portraits in the map art, improve the style transfer method, and transfer the input photo with the extracted portraits. 2. Through the deep convolutional neural network with different loss function, we combine the portrait and the map to produce map art. With the same method, we first classify the values ​​of each image, find the position of the portrait, and then train with the map art data to remove the portraits in the map art to get the restored background map. Our result not only can be made on new maps, but also on the map of the original map art. Since we can't get the correct map in the map art, and also there is not a lot of portrait image data of the same style as the map art portrait, in the training process, we find similar style maps and portraits, and combine them to be our training data. In addition, since the results are not optimal, we will optimize the results to achieve the best results through improved style transfer.

[1] L. A. Gatys, A. S. Ecker, and M. Bethge. Image style transfer using convolutional neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2414–2423, 2016.

[2] K. Simonyan and A. Zisserman. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014.

[3] F. Luan, S. Paris, E. Shechtman, and K. Bala. Deep Photo Style Transfer. arXiv preprint arXiv:1703.07511, 2017.

[4] C. Li and M. Wand. Combining markov random fields and convolutional neural networks for image synthesis. arXiv preprint arXiv:1601.04589, 2016.

[5] F. Luan, S. Paris, E. Shechtman, and K. Bala. Deep Painterly Harmonization. arXiv preprint arXiv:1804.03189, 2018.

[6] K. He, X. Zhang, S. Ren, and J. Sun. Deep Residual Learning for Image Recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016.

[7] C. Li, X. Liu, and T. Wong. Deep Extraction of Manga Structural Lines. ACM Transactions on Graphics (TOG), 2017. Article No. 117.

[8] O. Ronneberger, P. Fischer, and T. Brox. U-Net: Convolutional Networks for Biomedical Image Segmentation. arXiv preprint arXiv:1505.04597, 2015.

[9] S. Ioffe, and C. Szegedy. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. arXiv preprint arXiv:1502.03167, 2015.

[10] G. Liu, F. A. Reda, K. J. Shih, T.-C. Wang, A. Tao, and B. Catanzaro. Image Inpainting for Irregular Holes Using Partial Convolutions. arXiv preprint arXiv:1804.07723, 2018.

[11] Y. Wu, H. Liu, J. Li, and Y. Fu. Deep Face Recognition with Center Invariant Loss. Thematic Workshops '17 Proceedings of the on Thematic Workshops of ACM Multimedia 2017 Pages 408-414.

[12] K. Janocha, and W. M. Czarnecki. On Loss Functions for Deep Neural Networks in Classification. arXiv preprint arXiv:1702.05659, 2017.

[13] Y. Wen, K. Zhang, Z. Li, and Y. Chiao. A Discriminative Feature Learning Approach for Deep Face Recognition. In European Conference on Computer Vision. Springer, 499–515. 2016.

[14] Antonin Chambolle, Vicent Caselles, Matteo Novaga, Daniel Cremers, Thomas Pock. An introduction to Total Variation for Image Analysis. 2009. <hal-00437581>