Jack Hu

https://github.com/microsoft/SoftTeacher/issues/92

You can look at this one for reference.

一般半监督的时候标注数据比较少，会随机筛选5次（STAC工作的设定，被大家沿用）标注数据，然后以此制作5折标注-未标注的数据集，多次试验得到一个稳定结果，算均值方差啥的。如果你不需要多次验证来确保结果的可靠和稳定性，那么可以这么做。比如仅仅尝试一下，或者在某一折上面做消融实验。

Very good and important question. I'm curious about the relationship between weight setting and the final mAP performance. This can also benefit my work. I comment here simply to get...

可以的。做过了没问题。

我做消融的时候也没有遇到。。。。你的在有标签的数据上面正常吗，可以用训练出来的权重inference一下试试。。。如果即时训练轮数很大了也不正常的话那应该就是loss function等设计的问题了，如果正常的话那就挺奇怪的了。。。刚开始训练的时候没有伪标签是挺正常的

Same problem with you.Have you solved this?

Well, in my project of inpainting for medical images, reclustering is essential for greyscale medical images. Maybe code sharing is not essential, but the meaning of each element in kmeans_centers.npy...

I have rechecked the code and found the key to my question. Acually, the normalization method does not count too much. The code here define a correct relationship between the...

The kmeans npy is not in the range of [0, 1] as the annotation when I load it, please check whether there is mistake here. From the conversion rule, if...