Comments (4)
I think that's expected. Examples in *_train and *_val are fully-sampled and don't have the acceleration
parameter, while examples in *_test have undersampled k-space and the acceleration
parameter specifies the degree of undersampling (either 4 or 8). All examples have the acquisition
parameter, which specifies whether the image is fat-suppressed or not.
I'd have to ask @anuroopsriram, but it looks to me like the evaluate.py script is not intended to be run by participants of the fastMRI challenge. Rather it is included to show how submissions are evaluated on the server.
from fastmri.
Thank you for your response.
it looks to me like the evaluate.py script is not intended to be run by participants of the fastMRI challenge.
If so, it is better to delete description related with evaluate.py from Github.
For example, in https://github.com/facebookresearch/fastMRI/tree/master/models/unet, I find the following description.
The outputs will be saved to reconstructions_val. To evaluate the results, run:
python common/evaluate.py --target-path TARGET_DATA --predictions-path reconstructions_val --challenge CHALLENGE.
According to the description, I think that evaluate.py is intended to be run by participants of the fastMRI challenge.
By the way, I would like to reproduce the results of paper (https://arxiv.org/abs/1811.08839). Without evaluate.py and the attribute of acceleration, I think that Tables 4-7 cannot be reproduced by participants of the fastMRI challenge.
How do I reproduce the Tables 4-7 by using resources of the fastMRI challenge?
from fastmri.
You can run the evaluate.py
script on validation data without the --acceleration
argument. I have removed that argument from the code to remove the confusion.
For the tables in the paper, I used 4x or 8x sampling on the full validation data to get better estimates of the metrics. So, when you run run_model.py
, use --accelerations 4 --center-fractions 0.08
for 4x or --accelerations 8 --center-fractions 0.04
for 8x. Then run evaluate.py
on each of the outputs.
from fastmri.
from fastmri.
Related Issues (20)
- minor problem on `fastMRI_tutorial.ipynb` HOT 1
- knee data
- ValueError: too many values to unpack (expected 2) HOT 1
- Potential PyTorch Lightning dependency incompatibilities HOT 4
- Allow the data objects (module and dataset) to filter on contrast HOT 3
- Error related to raw_sample_filter in _create_data_loader HOT 3
- torch.eig is deprecated for a long time and is being removed HOT 4
- Error related to raw_sample_filter in _create_data_loader
- Problems with AnnotatedSliceDataset class HOT 3
- Unable to view public leaderboard submissions on the website HOT 2
- train_unet_demo produces wrong image HOT 1
- Unable to submit results to brain track HOT 1
- Unable to view public leaderboard submissions on the website HOT 2
- Tiny fastMRI data split proposal HOT 1
- A upgrade to PytorchLightning 2.0 and PyTorch 2.0 HOT 2
- Wrong annotations in AnnotatedSliceDataset HOT 5
- Training abruptly crashes on single GPU HOT 5
- Issue with argparse in PyTorch Lightning v2 HOT 1
- conda-forge build of fastmri? HOT 6
- Corrupted file for brain DICOM data HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from fastmri.