Hi, when i set the sr_scale from 4 to 2, there would be an error: RuntimeError: Caught RuntimeError in replica 0 on device 4. RuntimeError: cuDNN error: CUDNN_STATUS_NOT_INITIALIZED

It is in MASA_arch.py", line 122, fea_L1 = self.blk_L1(self.act(self.conv_L1(x))) I wonder what is happening

Dear author. The paper says that "The encoder consists of three building blocks – the second and third blocks halve the size of the feature maps with stride 2". However, the code's param "n_blks = [4, 4, 4]" in class MASA indicates that the second and third blocks have 2^4 times downsampling by stride 2, which is inconsistent with the paper. So, how to explain that?

Hi, I found this part of code can be optimized. When idx_x1.size(0) is a big number, the loop takes a lot of time. https://github.com/dvlab-research/MASA-SR/blob/9f0cccb71beafa764dadc5984fc8751cc0740c79/models/archs/MASA_arch.py#L343-L350 And here is my reimplementation

ind_y = idx_y1.view(-1, 1).repeat(1, diameter_x * s).view(-1)
ind_x = idx_x1.unsqueeze(-1).repeat(1, 1, diameter_y * s).permute(0, 2, 1).contiguous().view(-1)

Hi!

First of all I have to say that I am quite impressed by the possibility your current work has shown. However, after downloading the test data that you linked on your repo (CUFED5) and running the shell script for testing it using both the pre-trained models made available , I have been quite disappointed by the accompanying results :

(the reference images have been scaled to just fit into the presentation but I think they are of 500x332 resolution)

I have not looked too much into the test script that you have provided but I did look at the architecture you proposed in the paper and here are some observations/comments/questions I have:

1. shouldn't the Reference Image Resolution be 4 times the LR ? (this does not seem to be the case with the data provided)

2. currently upon running the shell script for test, the Resolution seems to be the same as the input, so is the shared implementation more like detail enhancement rather than SR ?

3. The input and Reference look to be of the same quality, so technically I would not be expecting massive improvements in the SR result, however if you could do something like DSLR based reference on a phones camera, it would be quite novel and interesting experiment. (This is the goal I hope to achieve using your methodology)

4. More theoretically, what kind of results would you expect when the input and reference are completely mismatched ?

Extremely sorry for the many naive questions that I might have asked but I hope to hear your views on the same!

Hello, Author. When I am testing with the code, I get the following error message：

PS D:\pythonProject\7_4\MASA-SR-main\MASA-SR-main> python test.py --resume './pretrained_weights/masa.pth' --testset TestSet_multi --name masa_TestSet_multi Traceback (most recent call last): File "test.py", line 95, in main() File "test.py", line 76, in main init_dist() File "D:\pythonProject\7_4\MASA-SR-main\MASA-SR-main\train.py", line 22, in init_dist rank = int(os.environ['RANK']) File "D:\python3.7\lib\os.py", line 681, in getitem raise KeyError(key) from None KeyError: 'RANK'

I hope you can reply to me, thanks.

Thanks for your great work!

I have some questions about definition of variables in MASA_arch.py. In forward of class MASA,

1. What do k_x and k_y mean, different from self.lr_block_size ?
2. Does diameter_x means the horizontal block size in Ref_LR ?
3. If k_x and k_y mean block_size, why using kernel_size of (k_x+2, k_y+2) when unfolding LR features, which is in line 374.

Looking forward to your reply and thanks in advance!

Thank you very much for your code and contribution, but I have some confusion, I use the data and weights you provided, and the result is like this, want to ask if there is a problem with this

RefSR/MASA-SR/models/archs/MASA_arch.py", line 435, in forward warp_ref_patches_x1 = warp_ref_patches_x1.view(N, C, H, W) RuntimeError: shape '[1, 64, 228, 228]' is invalid for input of size 3211264

it keeps making error..

Do you know why this occurs?

Absolutely amazing work, thanks for sharing. Could you please share with me where did you get the SUN80 dataset? I am having trouble finding it. Thanks in advance.