SR3:Image Super-Resolution via Iterative Refinement（零基础解读基于diffusion的超分网络）

Image Super-Resolution via Iterative Refinement

摘要： 本文提出来一种通过重复精细化操作来实现图像超分辨率的SR3模型。SR3将噪扩散概率模型嫁接至在图像到图像的翻译任务中，通过随机迭代去噪实现图像超分辨率。通过训练一个用于在各种水平噪声上去噪的U-Net架构，逐步的对完全高斯噪声进行逐步精细化最终得到输出图像。不论是人脸图像还是自然图像，SR3都在不同倍率的超分辨率任务上展示出超强的性能。我们在*We conduct human evaluation on a standard 8x face super-resolution task on CelebA-HQ for which SR3 achieves a fool rate close to 50%, suggesting photo-realistic outputs, while GAN baselines do not exceed a fool rate of 34%. We evaluate SR3 on a 4x super-resolution task on ImageNet, where SR3 outperforms baselines in human evaluation and classification accuracy of a ResNet-50 classifier trained on high-resolution images. We further show the effectiveness of SR3 in cascaded image generation, where a generative model is chained with super-resolution models to synthesize high-resolution images with competitive FID scores on the class-conditional 256x256 ImageNet generation challenge.(结果展示，与超分无关) *

简介：