search

uci-cbcl / NoduleNet

[MICCAI' 19] NoduleNet: Decoupled False Positive Reduction for Pulmonary Nodule Detection and Segmentation
Other
185 stars 57 forks source link

Is it possible to share a trained model? #11

Closed wenyalintw closed 4 years ago

wenyalintw commented 4 years ago

Hello developer! First of all, thanks for sharing your code, I truly appreciate your generosity. I'm trying to reproduce your work lately but not success yet. I'm wonder if you could share a trained model? Definitely for non-commercial use only.

If sharing through email is preferred, below is my email account, thanks a lot anyway! wenyalintw@gmail.com

wenyalintw commented 4 years ago

I train the model on one GeForce GTX 1060 6GB, which could only handle batchsize=2, not sure if this is why I couldn't train the model successfully.

For who has trouble installing build/box, I change all "long" to "int64_t" in nms.cpp to install it successfully on win10. (still lots of warning about data type when installing, not sure if it is important.)

tanghaotommy commented 4 years ago

Sorry for the delay. I will share a pertained weight shortly

tanghaotommy commented 4 years ago

https://drive.google.com/file/d/1B2aB_4HTw5OEe8jH6XPoQYRJw4hGuDXD/view?usp=sharing Please let me know if it does not work

valbertoenoc commented 4 years ago

@wenyalintw Could you please let us know how did you go about compiling build/box on windows? Thanks!