Cloud-Liu

Results 8 comments of Cloud-Liu

The test effect of the data set is much worse than the indicators reported in the original paper，How to solve it？

Hello, I have the same problem, have you solved it now?

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

> Hi, have you got the answer? 我觉得应该是用的验证集跑出的结果。他的论文中只写了将2594张图片分成0.8的训练集和0.2的验证集，代码中我也只看到分了训练集和验证集（你可以再详细看下）。所以他的表里的结果也就是验证集的结果指标。没有测试集的事。

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

跑了他的代码，能跑到0.9025和0.8290。我用他的代码，换成Unet结构跑到了0.8699和0.7805

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

你的IOU值怎么那么高，

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

不客气

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

我设置了随机种子数；设置的UNeXt-L：通道数分别是 32 64 128 256 512；或者你的batchsize增大一下

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

我觉得随机种子有点影响， ![image](https://user-images.githubusercontent.com/64886294/221349033-02f3721d-3134-464f-96b1-2faf0ad259c5.png) 是arch里边的通道数。

the IOU or DICE results on ISIC2018 dataset in the paper table, use valid dataset or the official test dataset?

我是直接改的