Image-Super-Resolution-via-Iterative-Refinement
Image-Super-Resolution-via-Iterative-Refinement copied to clipboard
Janspiry
Unofficial implementation of Image Super-Resolution via Iterative Refinement by Pytorch
First of all thanks a lot for this implementation. I am a student trying to understand the concept of Diffusion models and their applications. I am confused about the inputs...
hi,Thank you for the code.i have some question about the eval .Firstly, i see that you put sr image to the model testing.but i see the paper when sampleing that...
通过prepare.py得到的HR会极大地压缩质量？怎么改善呢？因为我本身的图像就是512*512大小的，经过这个脚本后，质量降低了很多，基本从300kb降低到40kb的程度
从我的val.log情况来看，为什么会出现PSNR值时高时低呢？不应该是随着迭代次数的增多，PSNR值稳步提升吗？ 以下是我的val.log记录： 找了很久都没发现原因，恳请告知！ 24-04-17 15:22:06.665 - INFO: psnr: 2.1367e+01 24-04-17 16:40:16.885 - INFO: psnr: 1.7173e+01 24-04-17 17:58:24.560 - INFO: psnr: 1.8179e+01 24-04-17 19:16:39.499 - INFO: psnr: 2.1871e+01 24-04-17 20:34:55.209 -...
When I only have low resolution images (16 * 16), how can I obtain the interpolated "high-resolution" images (SR) as input to UNet?
Why does this model also support unconditional generation?
I trained with SR3 using a dataset of thumbnails (128x128) with around 70,000 images (but after the downsampling process, the images were reduced to just 46,666). After that, I trained...
I was training on my own dataset for (128x128) --> (512x512) I have attached the train logs : [train.log](https://github.com/user-attachments/files/17504451/train.log) [val.log](https://github.com/user-attachments/files/17504453/val.log) The tensorboard visualizations look like this:  And the losses...
I trained and tested my own dataset. But my test results are some of them were good and some of them were just noise. My dataset was 580 size of...
Unexpected key(s) in state_dict: "feature_extractor.feature_extractor.0.weight", "feature_extractor.feature_extractor.0.bias", "feature_extractor.feature_extractor.2.weight", "feature_extractor.feature_extractor.2.bias", "feature_extractor.feature_extractor.5.weight", "feature_extractor.feature_extractor.5.bias", "feature_extractor.feature_extractor.7.weight", "feature_extractor.feature_extractor.7.bias", "feature_extractor.feature_extractor.10.weight", "feature_extractor.feature_extractor.10.bias".
