19tchoubou / anevrisme_segmentation

Aneurysm Segmentation with 3DUnet
Apache License 2.0
2 stars 0 forks source link

@Issue with the dataset used in this code #10

Open KuroTetsuya opened 11 months ago

KuroTetsuya commented 11 months ago

Hello, I am a new member of Github community. Thus, please forgive me if my question is inexperience or immature in anyway.

I am wondering as of which dataset was implemented in this code in order for me to try implementing and reproduce said result? In your readme file I cannot find the dataset information anywhere so I am kinda confused.

Wish you had a good day.

jumanamj commented 11 months ago

I believe you need to have your own dataset with extension .h5

Have you managed to properly install and import both classification_models_3D and segmentation_models_3D without any issues or errors?

KuroTetsuya commented 11 months ago

Thank you for your fast reply.

Yes, I have already installed and imported sad libraries without any issues or error. The cell within "train_3Dunet.ipynb" also print out "All libraries successfully imported".

Based on your comments, in order for me to implement this work, I should transform my dataset into .h5 extension or try to find a dataset that already in .h5 extension to test the code? The reason why I create the issue post was because I cannot find an example .h5 file originally from your work therefore I haven't actually run all the code yet.

Again, thank you very much for your reply.

jumanamj commented 11 months ago

When i run this cell

funky librairies for segmentation

from classification_models_3D.tfkeras import Classifiers import segmentation_models_3D as sm # pretrained models

import segmentation_models as sm

from patchify import patchify, unpatchify

i get this error:

ModuleNotFoundError Traceback (most recent call last) Cell In[4], line 2 1 # funky librairies for segmentation ----> 2 from classification_models_3D.tfkeras import Classifiers 5 import segmentation_models_3D as sm # pretrained models 6 #import segmentation_models as sm

File ~\anaconda3\Lib\site-packages\classification_models_3D\tfkeras.py:2 1 import tensorflow.keras as tfkeras ----> 2 from .models_factory import ModelsFactory 5 class TFKerasModelsFactory(ModelsFactory): 7 @staticmethod 8 def get_kwargs():

File ~\anaconda3\Lib\site-packages\classification_models_3D\models_factory.py:14 12 from .models import mobilenet as mb1 13 from .models import mobilenet_v2 as mb2 ---> 14 from .models import efficientnet as eff 15 from .models import efficientnet_v2 as eff2 16 from .models import convnext as cnext

File ~\anaconda3\Lib\site-packages\classification_models_3D\models\efficientnet.py:33 30 import math 32 from keras.applications import imagenet_utils ---> 33 from keras.engine import training 34 from keras.layers import VersionAwareLayers 35 from keras.utils import data_utils

ModuleNotFoundError: No module named 'keras.engine'

Do you have any idea why i get this? (just for info, i imported all other necessary libraries before this cell)

i am not a contributor in this repository, i am testing it out just like you are :) As for the file extension, i am actually using nifti and i changed the code accordingly. but in your case, try to use a dataset with .h5 extension

KuroTetsuya commented 11 months ago

@jumanamj: Hi, sorry for the misunderstand.

For the moment, I cannot find the cell you mentioned. However, I did copy the content of the cell and run on my device, which resulted in no error.

According to your error, the problem is due to the lack of module 'keras.engine'. My guess is that the keras library version and the efficientnet library version has a mismatch, thus during installation, the efficientnet cannot include the keras.engine module. Maybe you can try uninstall and reinstall both libraries for a quick check.

jumanamj commented 11 months ago

are you using jupyter notebook?

KuroTetsuya commented 11 months ago

@jumanamj: kinda similar, I had it run in VS Code with the Jupyter Notebook extension. Hope that you can solve your problem soon.

jumanamj commented 11 months ago

what version of python are you using?

KuroTetsuya commented 11 months ago

@jumanamj: I am currently using python 3.8, keras version 2.8.0 , classification_models_3D version 1.0.8, segmentation_models_3D version 1.0.5, and patchify version 0.2.3

sharafeldinkabashi commented 1 week ago

@jumanamj: I am currently using python 3.8, keras version 2.8.0 , classification_models_3D version 1.0.8, segmentation_models_3D version 1.0.5, and patchify version 0.2.3

Hello, May I ask if there is a source to the dataset that you used.