A quick question on R_Unet_v1

[YanhaoZhang]

Hi @vagr8 . I am reading R_Unet_v1.py, and have a quick question between line192-195:

 # use x5 to perform lstm
 if 'lstm_output' in locals():
    x6 = self.one_conv1(lstm_output)
    x5 = self.one_conv2(lstm_output)
    x5 = torch.cat((x5, x6), dim = 1)  

I understand lstm_output is the output after LSTM2, i.e, the grey arrow from figure 1 (https://github.com/vagr8/R_Unet/blob/master/runet_v1.jpg). May I ask why both x6 and x5 are calculated by lstm_output, or should one of them be the convolution by latent_feature?

Thanks a lot :).

[Michael-MuChienHsu]

My mistake, you are right, one should be latent_feature. Thanks!

[YanhaoZhang]

Haha, thanks a lot :)