search

ymli39 / DeepSEED-3D-ConvNets-for-Pulmonary-Nodule-Detection

DeepSEED: 3D Squeeze-and-Excitation Encoder-Decoder ConvNets for Pulmonary Nodule Detection
MIT License
106 stars 32 forks source link

The preprocess code does not generate all the label files #7

Closed Shrutika994 closed 4 years ago

Shrutika994 commented 5 years ago

Hello, I'm getting the following issue. I preprocessed the images but it happens not to generate a certain amount of label files which creates a problem while training. Would it be possible to look into this please? send to github

ymli39 commented 5 years ago

Did you check the name of the labels it generated instead?

Shrutika994 commented 5 years ago

Yes I checked the other labels it generated for the LUNA16 dataset but, while training, this problem seem to arise every time, so model can't be trained. So would it be possible that it takes the other label files as input for training and if it is possible could you please explain how.

ymli39 commented 5 years ago

Does the npy file exist in the directory or it is not generated. If the npy already exists, I recommand you sent a debug point there to see why it does not load.

Shrutika994 commented 5 years ago

The npy file is not generated.

ymli39 commented 5 years ago

I could generate all labels on my machine with no problem. I would recommend make a csv file that only contains label 360 to see the preprocessing results.

Shrutika994 commented 5 years ago

Okay I will try doing that. Thanks.

MjdMahasneh commented 3 years ago

@Shrutika994

The paper mentions 10-folds validation (which I presume, also means there is 10-fold training), my question is :

1-how is the 10-fold training done using the training script? 2-are these files (luna_train.npy and luna_test.npy) related to the 10-fold training? 3-how are these files generated?

Looking forward to hearing back from you.