siaaaaaa / cgc-net Goto Github PK

In the paper, it's stated that:

We extract patches with a size of 1792x1792

Do you expect the user to divide the images into patches or should I organize the folders without dividing the images into patches?

Thank you for sharing the code.

When I unzip both the fold1 and fold2 data from the TIA website I get an error suggesting it is corrupt. I tried to force unzip anyway but I only get raw images and no masks.

Could you link another source for this data please such as Google/Baidu Drive? Or give any suggestions you may have on how you unzipped it.

Thanks.

can't run with your code. What does this code mean?
for folder in self._processed_dir:
self.idxlist.extend([os.path.join(folder.split('/')[-1],f) for f in os.listdir(folder)])
print(self.idxlist)
self.feature_type = feature_type
（data.py line : about 163）

-----Auto Selecting Checkpoint Basing On "valid_dice" Through ">" Comparison
Selecting: ./media/vqdang/logs/v1.0/np_hv/model-29150.index
Having Following Statistics:
learning_rate: 9.999999747378752e-06
QueueInput/queue_size: 49.80662536621094
loss-bce: 0.11510444432497025
loss-dice: 0.1876249760389328
loss-mse: 0.019031917676329613
loss-msge: 0.3226739466190338
overall-loss: 0.6634671092033386
DataParallelInferenceRunner/QueueInput/queue_size: 49.90331268310547
valid_mse: 0.04642050700835553
valid_acc: 0.9463576229111165
valid_dice: 0.866723129027714
epoch_num: 50
global_step: 29150
[0910 15:18:37 @varmanip.py:215] Checkpoint path ./media/vqdang/logs/v1.0/np_hv/model-29150.index is auto-corrected to ./media/vqdang/logs/v1.0/np_hv/model-29150.
Traceback (most recent call last):
File "D:/Wonders_Code_files_Pycharm/20200907_cell_split/infer.py", line 183, in
inferer.run()
File "D:/Wonders_Code_files_Pycharm/20200907_cell_split/infer.py", line 149, in run
pred_config = PredictConfig(model = model_constructor(), session_init = get_model_loader(model_path), input_names = self.eval_inf_input_tensor_names, output_names = self.eval_inf_output_tensor_names)
File "C:\Users\Diting\anaconda3\envs\tensorflow1.10\lib\site-packages\tensorpack\predict\config.py", line 79, in init
self.input_signature = model.get_input_signature()
File "C:\Users\Diting\anaconda3\envs\tensorflow1.10\lib\site-packages\tensorpack\utils\argtools.py", line 192, in wrapper
value = func(*args, **kwargs)
File "C:\Users\Diting\anaconda3\envs\tensorflow1.10\lib\site-packages\tensorpack\train\model_desc.py", line 37, in get_input_signature
inputs = self.inputs()
File "C:\Users\Diting\anaconda3\envs\tensorflow1.10\lib\site-packages\tensorpack\train\model_desc.py", line 67, in inputs
raise NotImplementedError()
NotImplementedError

Process finished with exit code 1

The code for CIA-Net

Hello, this original picture does not have nuclei in it. How do you deal with this situation? thank you

Hi Yanning,

I cannot find the 'xtract-feature' package in pip installation which is refered in 'nuc_feature.py'.
Where can I install this package.

Thanks