LLDiffusion

LLDiffusion copied to clipboard

LOLv1 result

4

Dear author, this is an excellent work. While using the provided weights to perform inference on the LOLv1 dataset, I encountered results as shown in the attached image. I’d like...

whw2002

When I run test.py directly using the LOLv1 and LOLv2 datasets without resizing the images, and using the trained weights you provided, I encounter an error as follows. I would...

adpanda

Dear authors, Thank you for releasing your excellent work *"LLDiffusion: Learning Degradation Representations in Diffusion Models for Low-Light Image Enhancement"*. I am currently trying to reproduce your results, but the...

mousumibehera21

Hi there, Thank you for sharing your impressive work. I attempted to run your test scripts on lolv2-real using the default settings(batch-size=512 and grid_r=16) on a single 4090 GPU. However,...

AAAeray

代码

1

请问什么时候可以公布代码？论文里的图片恢复的和原文没有什么差别，因为我用扩散总是有一些小噪声，想看看作者的代码里面是如何解决这个的

Lairb

@vecbool @TaoWangzj thanks for sharing the work .. can u let us knw hwne the code and pre trained model is released Thanks in adavance

abhigoku10

你好，我已经读过你的论文，在实验设置里我看到了在训练时图片尺寸是128*128的，我之前也用过条件DDPM实验过低光照图像增强，但是，在推理时发现，因为要预先生成一张相同尺寸的纯高斯噪声图，图片尺寸要和训练时一样才行，也就是推理时图片也要是128*128。但是LOL数据集中图片尺寸是600*400，请问如何解决推理时图片尺寸与训练时不一致的问题。期待你的回复，谢谢。

superxianyuwang