DenseNet (Pretrained=Flase) in the Pytorch version

[c-yn]

When I read your code of Torch version, I found that the densenet used for the encoder is not pretrained on the Imagenet, which is different from the description in your paper. Thanks for your answering.