search

YuanXue1993 / SegAN

SegAN: Semantic Segmentation with Adversarial Learning
MIT License
182 stars 58 forks source link

'GlobalConvBlock' object has no attribute 'weight' #1

Closed vivek239 closed 6 years ago

vivek239 commented 6 years ago

I am trying to implement your code for medical image segmentation problem. When I tried to run the train.py file it always giving me an error 'GlobalConvBlock' object has no attribute 'weight'. Please try to resolve this issue.

YuanXue1993 commented 6 years ago

Will try this tomorrow. Could you give more details about the error?

vivek239 commented 6 years ago

when I fixed this above error model works , but during 1st epoch model freezed. Means not going to next epoch. Please try to fix all these error.

YuanXue1993 commented 6 years ago

I have fixed several bugs in train.py, now it's working well on my end. If you encounter out of memory error during the training, try to use a smaller batchsize. E.g., run with CUDA_VISIBLE_DEVICES=0 python train.py --cuda --batchSize 25

John1231983 commented 6 years ago

@YuanXue1993 : What is purpose for using GlobalConvBlock? Thanks