From 744c7af2ab4bf6622d71371f56d79dd5ac5b3abf Mon Sep 17 00:00:00 2001 From: Jiancheng Date: Tue, 2 Feb 2021 13:15:43 +0800 Subject: [PATCH 1/5] Update README.md --- README.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/README.md b/README.md index a937856..b0d01eb 100644 --- a/README.md +++ b/README.md @@ -76,6 +76,9 @@ python -m predict --image_dir --pred_dir In the [predict.py](predict.py), we adopt a post-processing procedure of [removing low-probability regions](predict.py#L18), [spine regions](predict.py#L24), and [small objects](predict.py#L48). This procedure leads to fewer false negatives. You may also skip the post-processing by setting `--postprocess False` in the command line argument and check the raw output. +***Note***: This project aims at a prototype for RibFrac Challenge; However, as the challenge data provider, we would like to avoid unintended data leakage. Therefore, we did not provide full details for the models, including those in both training and inference stages. Nevertheless, it is guaranteed that the performance in the EBioMedicine paper could be reproduced with this one-stage FracNet using 3D UNet as backbone. See more discussion in this [issue](https://github.com/M3DV/FracNet/issues/7). + + ### Evaluation To evaluate your prediction, run the following in command line: ```bash From fcfe06c804d33e37546ee1dce41c75c4cfed713f Mon Sep 17 00:00:00 2001 From: Jiancheng Date: Wed, 16 Jun 2021 10:43:15 +0200 Subject: [PATCH 2/5] Update README.md --- README.md | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index b0d01eb..849484b 100644 --- a/README.md +++ b/README.md @@ -15,7 +15,7 @@ For more details, please refer to our paper: *Liang Jin\*, [Jiancheng Yang](http://jiancheng-yang.com/)\*, [Kaiming Kuang](http://kaimingkuang.github.io/), [Bingbing Ni](https://scholar.google.com/citations?user=eUbmKwYAAAAJ), Yiyi Gao, Yingli Sun, Pan Gao, Weiling Ma, Mingyu Tan, Hui Kang, Jiajun Chen, Ming Li* -EBioMedicine ([DOI](https://doi.org/10.1016/j.ebiom.2020.103106)) +EBioMedicine, 2020 ([DOI](https://doi.org/10.1016/j.ebiom.2020.103106)) ## Code Structure * FracNet/ @@ -76,8 +76,9 @@ python -m predict --image_dir --pred_dir In the [predict.py](predict.py), we adopt a post-processing procedure of [removing low-probability regions](predict.py#L18), [spine regions](predict.py#L24), and [small objects](predict.py#L48). This procedure leads to fewer false negatives. You may also skip the post-processing by setting `--postprocess False` in the command line argument and check the raw output. -***Note***: This project aims at a prototype for RibFrac Challenge; However, as the challenge data provider, we would like to avoid unintended data leakage. Therefore, we did not provide full details for the models, including those in both training and inference stages. Nevertheless, it is guaranteed that the performance in the EBioMedicine paper could be reproduced with this one-stage FracNet using 3D UNet as backbone. See more discussion in this [issue](https://github.com/M3DV/FracNet/issues/7). +***Note 1***: This project aims at a prototype for RibFrac Challenge; However, as the challenge data provider, we would like to avoid unintended data leakage. Therefore, we did **NOT** provide all details for the models, including those in both training and inference stages. Nevertheless, it is guaranteed that the performance in the EBioMedicine'20 paper could be reproduced with this one-stage FracNet using 3D UNet as backbone, but also note that 1) it is trained with challenge data and extra in-house data, 2) there is heavy pre-processing and post-processing (including *rib segmentation*). See more discussion in this [issue](https://github.com/M3DV/FracNet/issues/7). +***Note 2***: Our paper on rib segmentation and centerline extraction has been recently accepted by MICCAI'21. Code and dataset will be available soon, please stay tuned! ### Evaluation To evaluate your prediction, run the following in command line: From fa817444e2835bd30224b23c5ca772fd5d2aa5d3 Mon Sep 17 00:00:00 2001 From: Shubham Agrawal <55156409+agrawalshubham01@users.noreply.github.com> Date: Wed, 6 Oct 2021 01:59:44 +0530 Subject: [PATCH 3/5] Update main.py --- main.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/main.py b/main.py index 3ed479b..18cf683 100644 --- a/main.py +++ b/main.py @@ -1,5 +1,7 @@ from functools import partial +import torch + import torch.nn as nn from fastai.basic_train import Learner from fastai.train import ShowGraph From a79a68ffac8c8b1d5073f1b4f19a038becb05372 Mon Sep 17 00:00:00 2001 From: Shubham Agrawal <55156409+agrawalshubham01@users.noreply.github.com> Date: Mon, 29 Nov 2021 15:41:14 +0530 Subject: [PATCH 4/5] Update main.py --- main.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/main.py b/main.py index 18cf683..42f4105 100644 --- a/main.py +++ b/main.py @@ -1,8 +1,7 @@ from functools import partial -import torch +from torch import save -import torch.nn as nn from fastai.basic_train import Learner from fastai.train import ShowGraph from fastai.data_block import DataBunch @@ -69,7 +68,7 @@ def main(args): ) if args.save_model: - torch.save(model.module.state_dict(), "./model_weights.pth") + save(model.module.state_dict(), "./model_weights.pth") if __name__ == "__main__": From bcccf983d12d6706e4e57af4e1ab4f95ace14d73 Mon Sep 17 00:00:00 2001 From: Jiancheng Date: Fri, 11 Feb 2022 14:36:30 +0100 Subject: [PATCH 5/5] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 849484b..3c2b54f 100644 --- a/README.md +++ b/README.md @@ -78,7 +78,7 @@ In the [predict.py](predict.py), we adopt a post-processing procedure of [removi ***Note 1***: This project aims at a prototype for RibFrac Challenge; However, as the challenge data provider, we would like to avoid unintended data leakage. Therefore, we did **NOT** provide all details for the models, including those in both training and inference stages. Nevertheless, it is guaranteed that the performance in the EBioMedicine'20 paper could be reproduced with this one-stage FracNet using 3D UNet as backbone, but also note that 1) it is trained with challenge data and extra in-house data, 2) there is heavy pre-processing and post-processing (including *rib segmentation*). See more discussion in this [issue](https://github.com/M3DV/FracNet/issues/7). -***Note 2***: Our paper on rib segmentation and centerline extraction has been recently accepted by MICCAI'21. Code and dataset will be available soon, please stay tuned! +***Note 2***: Our paper on rib segmentation and centerline extraction has been accepted by [MICCAI'21](https://arxiv.org/abs/2109.09521). Feel free to check it ;) ### Evaluation To evaluate your prediction, run the following in command line: