yinhaoz / denoising-fluorescence Goto Github PK

Hi!
Very nice work!
But when I run train_dncnn.py and benchmark.py in Google Colab, it reminds TypeError:

///
Traceback (most recent call last):
File "train_dncnn.py", line 133, in
patch_size=args.imsize, test_fov=19)
File "/content/denoising-fluorescence/denoising/utils/data_loader.py", line 412, in load_denoising
target_transform=target_transform, loader=pil_loader)
File "/content/denoising-fluorescence/denoising/utils/data_loader.py", line 137, in init
self.samples = self._gather_files()
File "/content/denoising-fluorescence/denoising/utils/data_loader.py", line 167, in _gather_files
if is_image_file(fname):
File "/content/denoising-fluorescence/denoising/utils/data_loader.py", line 28, in is_image_file
return has_file_allowed_extension(filename, IMG_EXTENSIONS)
File "/usr/local/lib/python3.6/dist-packages/torchvision/datasets/folder.py", line 20, in has_file_allowed_extension
return filename.lower().endswith(extensions)
TypeError: endswith first arg must be str or a tuple of str, not list
///

Do you know how to deal with it? Thank you so much.

xxxx> bash download_dataset.sh confocal
download_dataset.sh: line 33: [: missing `]'
Downloading dataset Confocal_BPAE_B...
--2021-09-13 14:03:54-- https://docs.google.com/uc?export=download&confirm=&id=1juaumcGn5QlFRXRQyrqfbZBhF7oX__iW
Resolving docs.google.com (docs.google.com)... 142.250.203.110, 2a00:1450:400a:801::200e
Connecting to docs.google.com (docs.google.com)|142.250.203.110|:443... connected.
HTTP request sent, awaiting response... 403 Forbidden
2021-09-13 14:04:00 ERROR 403: Forbidden.

Extracting files from ./dataset/Confocal_BPAE_B.tar...
tar: This does not look like a tar archive
tar: Exiting with failure status due to previous errors
Seems the tar file is wrong, download again...

Hello, I'm sorry to bother you.

But as far as I know, most images in the field of biomedicine are 16-bit tif images.

I tried some 16-bit images in the program, but because the PIL package cannot read 16-bit tif images, the program cannot run. After that, I converted the tif file into a 16-bit png format and sent it to test_example.py for testing. At this time, the image can be read, but the results are blank images. Then I converted the image to 8-bit png format, this time the program successfully completed the denoising task.

But 8-bit is missing a lot of information compared to 16-bit. So I would like to ask how to make this program run smoothly on 16-bit image set? And how to call the program when inputting 1024 * 1024 images?

thank you very much!

Hello, I'm sorry to bother you.

I retrained the DnCNN and Noise2Noise networks, tested the model, and got the denoising image in the test mix data set. I combined the images of three channels of focal BPAE into color images, but found that the resulting images are very different from the ground truth (color).

The gray images obtained from network training are all three-channel, while the gray images in the ground truth are single channel. I guess it may be related to this, but no solution has been found. How to deal with this?

Mentioned in the article: "This work is dedicated for fluorescence microscopy denoising where the images are corrupted by Poisson-Gaussian noise; in particular,Poisson noise, or shot noise, is the dominant noise source"

Due to the lack of dataset in x ray field（not in medical x-ray field），I want use this dataset for training. So could this dataset applied for x-ray deep denoising？the x-ray images are mainly corrupted by poisson noise.

In addition ,the dataset URL cannot be opened in mainland China,

hope for your reply!