I tried to follow the example code that was given, it gives the following error in Ubuntu 18.04, any help would be appreciated.
`ModuleNotFoundError Traceback (most recent call last)
in ()
4 from torch.autograd import Variable
5
----> 6 relaynet_model = torch.load('models/relaynet_good.model')

~/.local/lib/python3.6/site-packages/torch/serialization.py in load(f, map_location, pickle_module)
301 f = open(f, 'rb')
302 try:
--> 303 return _load(f, map_location, pickle_module)
304 finally:
305 if new_fd:

~/.local/lib/python3.6/site-packages/torch/serialization.py in _load(f, map_location, pickle_module)
467 unpickler = pickle_module.Unpickler(f)
468 unpickler.persistent_load = persistent_load
--> 469 result = unpickler.load()
470
471 deserialized_storage_keys = pickle_module.load(f)

ModuleNotFoundError: No module named 'networks.classifiers'
`

I've got the model working however it seems to only classify the top layer in an image (in black and white whereas paper had all layers being segmented in colour). Is this something that is being worked on? If not where can I contribute to bring this up to speed

In your paper and the original matlab code, you used Relu. However in this code you used Prelu

I noticed in your code that the test set is being used as the validation set. Could you please clarify if this approach is appropriate?

I get the following error code when trying to run the first few lines of code offline on my machine, since there is no test data set to run he model on there. We have some SD-OCT images which we could use there if we new the correct specifications to input to the model. We plan to cite your work and employ it in a further research paper:

File "/home/seanogor/Documents/Nandan/retinalwork/relaynet_pytorch/networks/data_utils.py", line 35, in get_imdb_data
Data = h5py.File('datasets/Data.h5', 'r')
File "/usr/local/lib/python3.5/dist-packages/h5py/_hl/files.py", line 312, in init
fid = make_fid(name, mode, userblock_size, fapl, swmr=swmr)
File "/usr/local/lib/python3.5/dist-packages/h5py/_hl/files.py", line 142, in make_fid
fid = h5f.open(name, flags, fapl=fapl)
File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper
File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper
File "h5py/h5f.pyx", line 78, in h5py.h5f.open
OSError: Unable to open file (unable to open file: name = 'datasets/Data.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

How should I download data.h5 file?

@abhi4ssj Could you open the 'Data.h5'? Thanks very much!!

Port to the new framework

If you can give me the code which can transfer raw data "Subject_*.mat" to datasets/Data.h5 ,It would be a great help . --a student who is very interested in the field of OCT segmentation.

Hi, thank you very much for sharing the code. I read your paper, it is a great paper. However, using this repo doesn't seem to be able to reproduce the good result in the paper. I also noticed that several details were not implemented, for example, the weigheted loss at the boundary.
I tried implemented myself, but get no luck. Even I used the UNet architecture (CM-UNet), I still can't reproduce the dice score in your paper. I got lower score on INL, OPL, and RPE.
Could you share more on the implementation details?

Thanks for your share. When I run this version (40ae1aa) code and I get the following error:
`/userfolder/relaynet_pytorch/relaynet_pytorch/net_api/losses.py in forward(self, output, target, weights, ignore_index)
60 encoded_target[mask] = 0
61 else:
---> 62 encoded_target.scatter_(1, target.unsqueeze(1), 1)
63
64 if weights is None:

RuntimeError: invalid argument 3: Index tensor must have same dimensions as output tensor at /pytorch/aten/src/THC/generic/THCTensorScatterGather.cu:289
`

And the input data's size is [NumData, 1, rows, cols], the corresponding label's size is [NumData, 1，rows, cols]. The target of our is to segment 10 class.
I read the code and don't understand these code:

`
relaynet_pytorch/relaynet_pytorch/net_api/losses.py

lin51：output = output.exp()

lin52：encoded_target = output.detach() * 0

line61：encoded_target.scatter_(1, target.unsqueeze(1), 1)

`

In the older versions, the implementation of DiceLoss is different from this version code. And the older vision as follows:

`
class DiceLoss(_Loss):

def forward(self, input, target):

    return 1 - dice_coeff(input, target)

`

So which is better ? or which is the right version for the "ReLayNet" paper ?
Thanks a lot!

Hello, I followed your instruction but confused about the datasets, I downloaded the datasets from the link (http://people.duke.edu/~sf59/Chiu_BOE_2014_dataset.htm) as mentioned in the paper, while the datasets consist of 610 annotated SD-OCT B-scan images of size 496x768 obtained from 10 patients suffering from DME (61 B-scans per patient). This is very different from the paper description, I think I had downloaded the wrong datasets, can you show the right link of the datasets download? thank you a lot!

RuntimeError: Traceback (most recent call last):
File "/home/sim/anaconda3/envs/py3/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 57, in _worker_loop
samples = collate_fn([dataset[i] for i in batch_indices])
File "/home/sim/anaconda3/envs/py3/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 57, in
samples = collate_fn([dataset[i] for i in batch_indices])
File "/home/sim/notebooks/relaynet_pytorch/relaynet_pytorch/data_utils.py", line 23, in getitem
weight = torch.from_numpy(weight)
RuntimeError: Trying to resize storage that is not resizable at /home/sim/install/pytorch/aten/src/TH/generic/THStorage.c:183

Based on your pytorch code, why is the dice I calculated greater than 1?

I'm getting the following error when trying to load a pretrained model.

model = torch.load('path/to/model/relaynet_epoch20.model')

Traceback (most recent call last):
File "C:\Users\gelaz\Anaconda3\lib\site-packages\IPython\core\interactiveshell.py", line 2862, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "", line 1, in
relaynet_model = torch.load(r'C:\Users\gelaz\PycharmProjects\VariationalOptimization\relaynet_pytorch\models\Exp01/relaynet_epoch20.model')
File "C:\Users\gelaz\Anaconda3\lib\site-packages\torch\serialization.py", line 367, in load
return _load(f, map_location, pickle_module)
File "C:\Users\gelaz\Anaconda3\lib\site-packages\torch\serialization.py", line 538, in _load
result = unpickler.load()
File "C:\Program Files\JetBrains\PyCharm Community Edition 2018.1.1\helpers\pydev_pydev_bundle\pydev_import_hook.py", line 20, in do_import
module = self._system_import(name, *args, **kwargs)
ModuleNotFoundError: No module named 'networks.relay_net'

When I try to run the test code I got
ModuleNotFoundError: No module named 'networks.relay_net'
It seems the "relaynet_pytorch" folder used to be called "networks" in your pretrained model.
Could you change the folder names back to "networks" to avoid such confusions? Thank you!

Great job! But I have a problem about the 2015_BOE_Chiu dataset, when I load the Subject_0x.mat file into matlab R2017a, all the variables are NaN. Do any one has idea about that?