zhuyiche / sfcn-opi Goto Github PK

Can you share the testing scripts for this project?

when i run the joint model, hava the error : ValueError: Operands could not be broadcast together with shapes (64, 64, 2) (64, 64, 5), did the model do well in the joint of det and cls, i am new to the keras code, so it's difficult to correct it, can you help me?

Hello, thank you for sharing the code. This problem occurs when I run your code. May be an environmental problem, my cuda is 9.0, cudnn is 7005, what is your cuda and cudnn? Or what do you think the problem is. Looking forward to your answer, thanks

When I run the joint model, I hava the error :ValueError: Operands could not be broadcast together with shapes (64, 64, 2) (64, 64, 5). Can you help me to slove this?

Thanks for the nice implementation!

I have a question about the tune_loss_weight function in 'model.py'.
I found that you used fixed weights to balance loss functions.
How do you find the proper weight values?
I want to apply this method to a new dataset, but it is very hard to search the weights by hands.
Is there any good strategy to find a proper loss weight?

I can't find the data_manager module, can you give any advice?

In the classification loss function, y_true is not used, there is something wrong with it.