Anime-InPainting

Anime-InPainting copied to clipboard

hahaha111111

About recall and precision

1

I found that the recall rate and accuracy rate are very low during my training. Is my training not enough? is this normal?I am using my own data set `512/512...

Demo122

我的图片和掩码都是1.2w，报错场景是这样的，有想过是图片的问题，但没能解决，求助一下博主，谢谢了 start training... Training epoch: 1 loading error: ./celeb/008621.png loading error: ./celeb/010481.png loading error: ./celeb/003145.png loading error: ./celeb/004050.png loading error: ./celeb/011577.png Traceback (most recent call last): File "train.py", line 4,...

fiveriddle

Pytorch版本

1

使用高版本的pytorch例如1.7.0会报错，使用低版本的pytorch又和CUDA不兼容，改怎么处理好

yankee1996

之前使用自己的数据集训练EdgeConnect模型，同样注意到了过久的训练时长。读源代码时只看到在loss.py中冻结了VGG19的部分参数，没有找到迁移学习有关的冻结网络层、调用预训练模型中参数等代码。 请问大佬具体是怎样实现的迁移学习？在项目代码中有写吗？感恩感恩~

dong0520

效果并不好

3

总是把眼睛花在鼻子上面 

primejava

运行不了/ I can't run the tool

6

多谢学长代码。我运行了工具，用的win10CPU，只用input和output两个窗口，可以显示 Image selected: G:\yzceig\Pictures\input.png，前面打印的也没问题。然后就用左键划黑线，但按‘n'键无反应，命令行光标一直在闪，按其他键也一样无反应。这可能是什么问题。

yzceig

为什么xiaogyu效果是这样的还是我的操作方法的问题呢？作者大哥看到的话请联系我谢谢了，3241644639 (qq)

fangxuanhao

i use the pre-trained net to run the tool .but these mistakes happened

1

RuntimeError: Error(s) in loading state_dict for EdgeGenerator: Missing key(s) in state_dict: "encoder.1.weight", "encoder.4.weight", "encoder.7.weight", "middle.0.conv_block.1.weigh t", "middle.0.conv_block.5.weight", "middle.1.conv_block.1.weight", "middle.1.conv_block.5.weight", "middle.2.conv_block.1.wei ght", "middle.2.conv_block.5.weight", "middle.3.conv_block.1.weight", "middle.3.conv_block.5.weight", "middle.4.conv_block.1.w eight", "middle.4.conv_block.5.weight", "middle.5.conv_block.1.weight", "middle.5.conv_block.5.weight", "middle.6.conv_block.1...

582751489

关于数据格式问题

1

我在用paris数据集进行model 1 的训练时出现了一些错误 我的数据集尺寸为256*256 mask数据集也为256*256，下面是我的错误截图，请问您项目中的初始数据集尺寸与mask尺寸为多少呢？ 

897486562