how to do the dataset?

[wxde]

Hi Fausto: I am a newer,I'm trying to use your 3D-caffe, and use your vnet , i don't know how do you make your dataset (such as should I transform it to lmdb?) , could you please tell me the founction of each flie.py from xu

[faustomilletari]

Hello,

for Vnet you have just to have two folder, one for training and the other for testing with mhd files which correspond to images and binary segmentation masks. The filenames for the images can be whatever but the filenames for the segmentations must be the same name that you used for the corresponding training image followed by ‘_segmentation’

so if you have in your ./training folder the files a.mhd b.mhd you must also have a_segmentation.mhd b_segmentation.mhd

because of the mhd format you will also have some .raw files, namely:

a.raw b.raw a_segmentation.raw b_segmentation.raw

have a look at the code to understand more, or use the promise12 dataset to try out.

Fausto

On Apr 8, 2017, at 3:27 AM, wxde notifications@github.com wrote:

Hi Fausto: I am a newer,I'm trying to use your 3D-caffe, and use your vnet , i don't know how do you make your dataset (such as should I transform it to lmdb?) , could you please tell me the founction of each flie.py from xu

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/25, or mute the thread https://github.com/notifications/unsubscribe-auth/AMtsvkVNbNPaN8TRizS6db0MEG4kCGI2ks5rtzbhgaJpZM4M3n9d.

[wxde]

Fausto , thank you fou your reply I have a try by your suggestion from xu