Miao Zheng

@linfangjian01 Pay attention to this issue

what does the output result mean? `train_set` is an object that is pytorch dataset class

would you mind providing the full config and where are the defalut locations?

I think the simplest solution is to use `class_weights` in loss to reduce specific category effect

Yes you are right

@ice-tong please have a look

It's a good question that the different pretrained models will help the results on ade20k, and many excellent scientists are trying to figure it. In fact, we are planning to...

Thanks for your reminder.

It is a very good proposal, and also in mmseg's backlog, but we suffer from a shortage of manpower, so 3d segmentation might not be supported in 2022. Any collaborations...

I suggest you add a new data transform what you need in https://github.com/open-mmlab/mmsegmentation/blob/master/mmseg/datasets/pipelines/transforms.py and make you own dataset config file. Please refer this tutorial to understand and extend mmseg data...