【Bringing Old Photos Back to Life】Mapping(mapping_net)映射网络
发布日期:2021-07-01 02:26:48 浏览次数:2 分类:技术文章

本文共 1281 字,大约阅读时间需要 4 分钟。

1.With the emergence of deep learning, one can address a variety of low-level image restoration problems [
5
,
6
,
7
,
8
,
9
,
10
,
11
,
12
] by exploiting the powerful representation capability of convolutional neural networks,
i.e., learning the mapping for a specific task from
a large amount of synthetic images.
 

随着深度学习的出现,人们可以利用卷积神经网络强大的表示能力,即学习来解决各种低级图像恢复问题[5, 6, 7, 8, 9, 10, 11, 12] 来自大量合成图像的特定任务的映射 

 

2.The mapping between the two latent spaces is then learned with the synthetic image pairs, which restores the corrupted images to clean ones.

然后使用合成图像对学习两个潜在空间之间的映射,将损坏的图像恢复为干净的图像。 

 

3.Restoration via latent space translation :In order to mitigate the domain gap, we formulate the old photo restoration as an image translation problem, where we treat clean images and old photos as images from distinct domains and we wish to learn the mapping in between.

通过潜在空间转换进行恢复:为了减少域间隙,我们将旧照片恢复制定为图像转换问题,我们将干净的图像和旧照片视为来自不同领域的图像,我们希望学习它们之间的映射。 

 

4.Then,we learn the mapping that restores the corrupted images to clean ones in the latent space.

然后,我们学习将损坏的图像恢复到潜在空间中干净的图像的映射。

 

5.VAEs with two-stage training (VAEs-TS): the two VAEs are first trained separately and the latent mapping is learned thereafter with the two VAEs (not fixed);

两阶段训练的VAE(VAEs-TS):首先分别训练两个VAE,然后用两个VAE学习潜在映射(不固定);

 

 

 

 

 

 

 

转载地址:https://mapengsen.blog.csdn.net/article/details/117981667 如侵犯您的版权,请留言回复原文章的地址,我们会给您删除此文章,给您带来不便请您谅解!

上一篇:【Bringing Old Photos Back to Life】How to train?如何训练
下一篇:【Bringing Old Photos Back to Life】nonlocal block(非局部块)==>针对结构化缺陷,利用全局上下文来恢复潜在特征

发表评论

最新留言

做的很好,不错不错
[***.243.131.199]2024年04月14日 12时33分25秒