onurboyar / retinal-vessel-segmentation

Retinal Vessel Segmentation using U-Net architecture. DRIVE and STARE datasets are used.

• 👋 Hi, I’m @onurboyar
• 👀 I’m researcher at RIKEN and Ph.D. Student at Nagoya University, Japan.
• My research interests are: Latent Space Bayesian Optimization, Generative Models for Sequence data, MCMC
• 📫 Personal Website: onurboyar.com

retinal-vessel-segmentation's Issues

Test only 27 images

@onurboyar

Thank you for your code, it is beneficial to me.

I have my private dataset, and my U-net code runs correctly (train and test) on the full dataset.
I want to test the model on only 27 images. When I execute evaluate function, I meet this error :

**InvalidArgumentError: 2 root error(s) found.
(0) Invalid argument: slice index 1 of dimension 2 out of bounds.
[[{{node strided_slice_1}}]]
[[IteratorGetNext]]
[[IteratorGetNext/_4]]
(1) Invalid argument: slice index 1 of dimension 2 out of bounds.
[[{{node strided_slice_1}}]]
[[IteratorGetNext]]
0 successful operations.
0 derived errors ignored. [Op:__inference_test_function_12429]

Function call stack:
test_function -> test_function**

Can you please help me?
thank you

System requirements

Hi,

What is the minimum system requirements to run this code?

Thanks

When I run train_drive, it has the problem"unable to open file /path/to/ckpts.hdf5",how can I get it?

If I want to run this code from scratch, how should I do?

Pretrained Weights

Thanks for sharing your work.
Please share your Pretrained Weights for this work for further analysis, as it would help me and verify your results.
I really appreciate it if you answer these questions. Thanks very much.

I got empty png result

Hi,

My result are empty png as below. I used drive dataset and I convert all images to png before traninig.

How can I fix it?

Thank you