search

xinwucwp / faultSeg

Using synthetic datasets to train an end-to-end CNN for 3D fault segmentation (We are working on an improved version!)
206 stars 130 forks source link

train the model #16

Closed super-wu12138 closed 2 years ago

super-wu12138 commented 2 years ago

Hi, Dr Wu. Thank you very much for this awesome research！ I have run the code successfully, but it seems that I can't get "pretrained_model.hdf5" by training the model. Could you please tell me How to train the model will get "pretrained_model.hdf5". Thanks

llmpass commented 2 years ago

hi, the google drive link share in this repo still alive: https://drive.google.com/drive/folders/1q8sAoLJgbhYHRubzyqMi9KkTeZWXWtNd I had just double checked it.

super-wu12138 commented 2 years ago

Okay, thanks for your comments. I have download "pretrained_model.hdf5" before successfully, but my doubt is how can I get the "pretrained_model.hdf5" by running code or train the model “Unet3” ，not download via the he google drive link.

llmpass commented 2 years ago

I think just following the setting in the paper will more or less achieve a very similar (off course not 100% the same) model.