让我们看看艺术家眼里的世界吧！ 我们训练了一个深度学习网络学习多个名家的画风，并且可以让它模仿各种风格来”画“新的图片。 请看一下我们的演示视频，优酷地址：用艺术家的眼光看世界-在线播放-优酷网，视频高清在线观看 （YouTube 地址 https://www.youtube.com/watch?v=oy6pWNWBt4Y）
这是基于 ICCV2017 投稿的论文 (Multi-style Generative Network for Real-time Transfer)，在这里简单说一下原理，因为笔者是到美国之后才接触的科研，所以术语使用不当之处还请多包涵（实在是捉急，请多指正）。在此论文之前，已经有比较优秀的工作(Ulyanov和Johnson)实现了实时的(real-time)风格转换(Style Transfer)，他们的工作主要是受Gatys工作的启发，用相同的损失函数训练一个生成网络，从而将复杂的运算负担(burden)放在了训练的过程在，实现了实时的画风转换。但是这种算法的可扩展性(scalability)很局限，需要对每一种风格，训练不同的生产网络。
那么是什么限制了已有的算法，使其不能够生成多样的风格呢？根据Gatys的理论，风格和纹理信息可以用特征的统计信息 feature statistics来描述，已有的生产网络的框架下，这个风格图片的统计信息是通过损失函数隐含地(implicitly)传达给生产网络，而不是直接让生成网络来明确地(explicitly)与风格图片的特征相匹配(match)。本文提出了一个灵感层(Inspiration Layer)，保留原有特征的内容信息，并且匹配代表风格和纹理的统计信息。我们在不同的尺度下进行特征匹配，从而实现了多种风格的转换。我们将运算的负担留在了训练中，从而实现了实时的转换。
最后展示一下我们方法的效果，以下多样的图片全部由一个生成网络生成的：
论文《Multi-style Generative Network for Real-time Transfer》摘要：
Recent work in style transfer learns a feed-forward generative network to approximate the prior optimization-based approaches, resulting in real-time performance. However, these methods require training separate networks for different target styles which greatly limits the scalability. We introduce a Multi-style Generative Network (MSG-Net) with a novel Inspiration Layer, which retains the functionality of optimization-based approaches and has the fast speed of feed-forward networks. The proposed Inspiration Layer explicitly matches the feature statistics with the target styles at run time, which dramatically improves versatility of existing generative network, so that multiple styles can be realized within one network. The proposed MSG-Net matches image styles at multiple scales and puts the computational burden into the training. The learned generator is a compact feed-forward network that runs in real-time after training. Comparing to previous work, the proposed network can achieve fast style transfer with at least comparable quality using a single network. The experimental results have covered (but are not limited to) simultaneous training of twenty different styles in a single network. The complete software system and pre-trained models will be publicly available upon publication.
论文链接：
https://arxiv.org/abs/1703.06953
代码链接：
https://github.com/zhanghang1989/MSG-Net
知乎专栏：
https://zhuanlan.zhihu.com/p/25892708
原文链接：
http://weibo.com/1402400261/EBwAKnTlV?type=comment#_rnd1490523788150
