Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

如何在非人脸（有人脸的特征）数据集上训练出您样本中优秀的效果？ #59

Open
seoseven opened this issue Mar 6, 2023 · 6 comments
Open

如何在非人脸（有人脸的特征）数据集上训练出您样本中优秀的效果？ #59

seoseven opened this issue Mar 6, 2023 · 6 comments

Comments

@seoseven
Copy link

seoseven commented Mar 6, 2023

您好，DaGAN展示出的效果非常好，但当我运行自己的数据集时，效果却很差，
@seoseven
Copy link
Author

seoseven commented Mar 6, 2023

就像这样的图像

@seoseven seoseven closed this as completed Mar 6, 2023
@seoseven seoseven reopened this Mar 6, 2023
@harlanhong
Copy link
Owner

我看不到你的图像啊~

@seoseven
Copy link
Author

seoseven commented Mar 6, 2023

抱歉，打扰您了，我以为我成功上传了图片
image
seed6601
像这样，类似中国戏曲中的脸谱图像，例如我想将明星唱歌驱动视频里的动作迁移至此类图像上，结果是图像中眼睛嘴巴产生的形变很小，几乎没有。我是否需要重新训练一个基于此类图像的关键点侦测模型，还是说这种图像不可能实现像cartoon sample中如此好的效果，期待您的回复！谢谢

@harlanhong
Copy link
Owner

这个是数据之间的domain gap, 我用的那些cartoon sample还是具备了人脸的形态，跟训练集之间的gap还是较小的。你这种数据跟训练集的gap太大了

@seoseven
Copy link
Author

seoseven commented Mar 7, 2023

所以说，我只能通过去训练大量上述脸谱图像为主体内容的视频序列集合，这样得到的模型才有可能成功吗？还是说这样做其实也不行，因为目前的image animation方向还是针对人脸为主？

@harlanhong
Copy link
Owner

是的，如果需要效果好一些，就需要上述的数据进行训练

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@harlanhong @seoseven