xmindflow/DAEFormer

Verification of the details of the DAE-Former paper

ai-JIE opened this issue · 4 comments

ai-JIE commented

Hello Author, recently I have been doing research on medical image segmentation. For the dae article you wrote, I have the following doubts about the training strategy code for training isic data sets that you did not disclose, this part of the code is very important to me. When I tried to replicate it myself, it was much lower. Hope to get your reply.

amirhossein-kz commented

Hi @ai-JIE,

For information regarding the training and evaluation code for the skin dataset, please refer to this link.

ai-JIE commented

嗨，

有关皮肤数据集的训练和评估代码的信息，请参阅此链接。
According to your suggestion, I replicated the experiment of DAE-Foremer's skin dataset and obtained a Dice-score of 0.86. I'm shocked.

yangz9527 commented

嗨，
有关皮肤数据集的训练和评估代码的信息，请参阅此链接。
According to your suggestion, I replicated the experiment of DAE-Foremer's skin dataset and obtained a Dice-score of 0.86. I'm shocked.

Hello, may I ask how you reproduced the results on the ISIC dataset? I tried to reproduce the results on the ISIC dataset but failed

ZBWKHH commented

嗨，
有关皮肤数据集的训练和评估代码的信息，请参阅此链接。
According to your suggestion, I replicated the experiment of DAE-Foremer's skin dataset and obtained a Dice-score of 0.86. I'm shocked.

Hello, may I ask how you reproduced the results on the ISIC dataset? I tried to reproduce the results on the ISIC dataset but failed

When I used the ISIC2018 dataset, referring to the link provided by the author, DSC obtained 85.7