刘欣雨 comments

Results 7 comments of


                                            刘欣雨

The value Acc of the test result is nan, Dice and Iou are 0

> > I'm sorry that from the info you provided, I can't locate the problem. I think you can evaluate the test set and validation set separately to check if...

The value Acc of the test result is nan, Dice and Iou are 0

It might be due to too few images in the test or validation set, and insufficient training of the network in the early stages.my val just have 1 image.so..

Segformer for Binary Segmentation task - NaN loss

I met the same issue!!!

Segformer for Binary Segmentation task - NaN loss

It might be due to too few images in the test or validation set, and insufficient training of the network in the early stages.my val just have 1 image.so..

customized rawpy directory

I also encountered the problem of fine-tuning on other datasets and found that the effects of NEFNet and Restormer are not as good as the effects of Unet. Could you...

尝试使用了LED + NAFNet 与 LED + Restormer 效果都不如Unet好

> 您好，可以提供下您具体的数据以及设置吗？您好,我使用的是 ELD 数据集具体的配置文件为默认的文件在 LED 提供的基本配置上没有做任何改动

NAFNet版本LED疑问

I also encountered the problem of fine-tuning on other datasets and found that the effects of NEFNet and Restormer are not as good as the effects of Unet. Could you...