dattran2346 / chestx-ray-14 Goto Github PK

Can you please upload best.h5 file? I am trying to implement your code and I don't have the required GPU to train the dataset.

I am trying to reproduce this model you have implemented. However, being a fairly novice user, I am having trouble doing so.
Can you provide more details as to how to run and build your code (on Jupyter notebook)?

Hey
I am not able to understand how masks generated by UNET are used during the training?
apart from cropping the Xray, is there anything else being done using masks?

Hi
Thanks for yur work. There is an issue though.
in utils.blend_segmentation, in the very first line, it is h, w = image.size
but it should be w, h = image.size

Hi @dattran2346,
Can you please clarify a few questions:

1. What does the cut_all.pickle file contain?
2. For the training of the Unet model, the labels for the data is the binary image containing a mask of both lungs right?
3. Can you please guide us a bit more as to how to test your Unet model on another data set?
4. Also, if possible, can you give a brief description of what each of your .ipynb file is for?

@AleenaAzam

When I use the LFS dropdown unet.h5 file, I get the error that "batch response: this repository is over its data quota. Account response for LFS bandwidth should purchase more data packs to restore access."
I would like to further study your code, if you can send unet.h5 model to me by email. The email address is [email protected]. Thank you very much.