hello,can you add the code of generating result.csv in your predict.py?if not ,that evaluation will not pass.

M3DV / FracNet

[eBioMedicine] Deep-learning-assisted detection and segmentation of rib fractures from CT scans: Development and validation of FracNet

https://m3dv.github.io/FracNet/

Apache License 2.0

51 stars 25 forks source link

hello,can you add the code of generating result.csv in your predict.py?if not ,that evaluation will not pass. #2

Closed JXQI closed 3 years ago

duducheng commented 3 years ago

Sorry for this mistake, we will fix it soon.

JXQI commented 3 years ago

At the same time，I think your main.py can add the code of saving model after training!

example: torch.save(model.state_dict(),"./models/3dUnet.pth")

then,in predict.py : "model = nn.DataParallel(model).cuda()" should follow the "model = UNet(1, 1, n=16)" do like this ,the code will not Interrupt！ just suggest!

kaimingkuang commented 3 years ago

New commits pushed. Now predict.py generates .nii and .csv files in alignment with the RibFrac challenge requirements.