Giters

orobix / retina-unet

Retina blood vessel segmentation with a convolutional neural network

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Prediction of a general fundus image from hospital

anitstanly opened this issue · comments

commented

Hi
I used another data for testing but it wont work , did not showing any result and got error as "IOError: Unable to open file (unable to open file: name = './DRIVE_datasets_training_testing/DRIVE_dataset_imgs_test.hdf5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)"
how can solve this? please help me

commented

Hi
I used another data for testing but it wont work , did not showing any result and got error as "IOError: Unable to open file (unable to open file: name = './DRIVE_datasets_training_testing/DRIVE_dataset_imgs_test.hdf5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)"
how can solve this? please help me

Have you solved it? I also fail as well as you……

commented

Helo
No, I couldnt solve this problem.......
Here in this code ,Stare result is provided. How did it generated using this code .......

commented

Hi,
for STARE dataset, you may have to change the first part of the configuration.txt file as follow:

[data paths]
path_local =  /content/drive/MyDrive/Dataset/STARE_datasets_training_testing/          # you have to put the correct path here
train_imgs_original = STARE_dataset_imgs_train.hdf5
train_groundTruth = STARE_dataset_groundTruth_train.hdf5
train_border_masks = STARE_dataset_borderMasks_train.hdf5
test_imgs_original = STARE_dataset_imgs_test.hdf5
test_groundTruth = STARE_dataset_groundTruth_test.hdf5
test_border_masks = STARE_dataset_borderMasks_test.hdf5

and for the code, you have to do Ctrl+H and replace DRIVE with STARE.
you also need to preprocess the dataset beforehand. for me, i rescaled all STARE images to be the same size as DRIVE images, and i created mask for every image to be exactly as in DRIVE dataset. finally don't forget to rename all the images in STARE to be as the images' names in DRIVE dataset.

Note: i used Colab Pro to execute the code as the RAM offered in free Colab is not sufficient to handle all 190000 patches.

Hope this helps.
Regards