2D multi-class classification tasks

[cdzheng2g]

I was wondering how I might adapt it for 2D multi-class classification tasks. please let me know what modifications might be needed?

[JCruan519]

Following the same approach as the skin disease segmentation dataset, modify the loss function to CeDiceLoss. This might work.

[ouycc]

按照与皮肤病分割数据集相同的方法，将损失函数修改为 CeDiceLoss。这可能会有效。

I used my own dataset with 20 classes and modified the loss function to CeDiceLoss in the same way as the skin lesion segmentation dataset. When training to the 30th epoch, I encountered the error 'Found input variables with inconsistent numbers of samples: [6029312, 120586240]'. Do I need to modify the CeDiceLoss function