henryccl / fpnet Goto Github PK

about ‘Source free domain adaptation for medical image segmentation with fourier style mining’
Hello, I'm sorry to bother you. I noticed that you have reproduced this paper, and I'm wondering if I can ask you a few questions.

Firstly, I would like to ask if you trained the source model by writing your own code or using DeepLabv2?

The second question is what you used for the pre training model in deeplabv2. Is it resnet101? I don't know why I reported an error when using it to train my source model. There was a loading problem
您好打扰您一下，我看到您复现了这篇论文，有几个问题不知道能否请教下您。
首先，我想问一下源模型您是自己写代码训练的吗，是用deeplabv2训练的吗？
第二个问题是您在deeplabv2中的预训练模型用的什么。是resnet101的吗，不知道为什么我在使用他去训练我的源模型时会报错加载有问题

作者你好，我在复现时遇到了问题。具体问题是有关你的网络结构在0.3bpp嵌入率及以下嵌入率时，存在不收敛的问题。我想咨询一下有关你的网络参数初始化的方法是什么？这是非常重要，它直接影响网络的收敛问题。