search

zaccharieramzi / fastmri-reproducible-benchmark

Try several methods for MRI reconstruction on the fastmri dataset. Home to the XPDNet, runner-up of the 2020 fastMRI challenge.
https://fastmri.org/leaderboards
MIT License
151 stars 50 forks source link

About fastmri_recon/config.py #116

Closed fharman closed 3 years ago

fharman commented 3 years ago

Hi,

I want to ask you about fastmri_recon/config.py. In test folder, 108 h5(AF4=50, AF8=58) files exist. But in the main code, as you see there are 120 files .

import os

FASTMRI_DATA_DIR = os.environ.get('FASTMRI_DATA_DIR', '/media/Zaccharie/UHRes/') OASIS_DATA_DIR = os.environ.get('OASIS_DATA_DIR', '/media/Zaccharie/UHRes/') LOGS_DIR = os.environ.get('LOGS_DIR', './') CHECKPOINTS_DIR = os.environ.get('CHECKPOINTS_DIR', './')

n_volumes_train = 973 n_volumes_val = 199 n_volumes_test = { 4: 60, 8: 60, }

Am i wrong about the number of test files?

Thank you for your patience and support.

Best Regards,

zaccharieramzi commented 3 years ago

Hi @fharman ,

You are not wrong, there are 108 test files. These 2 numbers are indeed not accurate but it's not too sever: they are used only in harmless places in the code (for tqdm logging). I will fix it in a coming PR.

I see that you are still using an old version of the code. As mentioned here, you could now safely use the current version with the dedicated evaluation script to reproduce the results of the paper.

fharman commented 3 years ago

Hi Zaccharie,

I want to ask you about the difference. Because the latest one, I remember you used (Glorot uniform(Xavier Initialization,), Rectified Adam for Optimizer, and not used kernel_initializer='he_normal') .But in the new one, there is a difference I realized. Why is there a big difference?

Thank you for your return,

Best regards,

Zaccharie Ramzi notifications@github.com, 27 Kas 2020 Cum, 00:03 tarihinde şunu yazdı:

Hi @fharman https://github.com/fharman ,

You are not wrong, there are 108 test files. These 2 numbers are indeed not accurate but it's not too sever: they are used only in harmless places in the code (for tqdm logging). I will fix it in a coming PR.

I see that you are still using an old version of the code. As mentioned here https://github.com/zaccharieramzi/fastmri-reproducible-benchmark/issues/115#issuecomment-734426749, you could now safely use the current version with the dedicated evaluation script https://github.com/zaccharieramzi/fastmri-reproducible-benchmark/blob/master/fastmri_recon/evaluate/scripts/paper_eval.py to reproduce the results of the paper.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/zaccharieramzi/fastmri-reproducible-benchmark/issues/116#issuecomment-734476605, or unsubscribe https://github.com/notifications/unsubscribe-auth/AO4IQ2WZZXT3LMK3PSMSVI3SR27APANCNFSM4UEDULGQ .

zaccharieramzi commented 3 years ago

@fharman I am not sure what you mean and I don't think it's related to this question. Please open a new issue if you have another question.

I closed this with the PR #117 being merged.