search

ai-med / relaynet_pytorch

Pytorch Implementation of retinal OCT Layer Segmentation (with trained models)
MIT License
87 stars 30 forks source link

data.h5 #9

Closed CJ0109 closed 6 years ago

CJ0109 commented 6 years ago

How should I download data.h5 file?

lz666win commented 6 years ago

the same question

abhi4ssj commented 6 years ago

Hi, the dataset is from Duke University (open-source) detailed in the ReLayNet paper. I converted to h5 format myself. You need to write your own DataLoader if your data is in some other format. Please note, I cannot share the h5 file as Duke University has the re-distribution rights.

lz666win commented 6 years ago

I found the link of Duke dataset,but there is only image data,not have label data.Do you make label all by yourself?

abhi4ssj commented 6 years ago

Hi, The labels are provided (from 2 experts). It is provided in a different format. For each A-scan, only the co-ordinates where transitions of layers occur is provided. You need to write a small script to convert it to image size. Best Abhijit

shayansiddiqui commented 6 years ago

Closing issue because of inactivity

ZeyuFu commented 5 years ago

Would you please give me some hints how to convert the these transition layers (876861) into an image? I have no idea how match these labels with the original image.