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.

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

梯度监督的作用似乎不是非常明显? #183

Open
Lucky-Jay opened this issue Feb 17, 2025 · 1 comment
Open

梯度监督的作用似乎不是非常明显? #183

Lucky-Jay opened this issue Feb 17, 2025 · 1 comment

Comments

@Lucky-Jay
Copy link

你好

我这边可视化了开源的BiRefNet-matting-epoch_100.pth和BiRefNet-portrait-epoch_150.pth两个抠图模型对梯度预测的结果图

结果如下:
BiRefNet-matting-epoch_100.pth:
Image
Image
Image
Image

          BiRefNet-portrait-epoch_150.pth:
          ![Image](https://github.com/user-attachments/assets/a022890d-7e31-43b6-83c3-ab2cb1d8232e)
          ![Image](https://github.com/user-attachments/assets/56448566-e6b3-432c-a542-03f8bdf6bb82)
          ![Image](https://github.com/user-attachments/assets/2ea0a847-4e56-4987-a376-fc31b6dc1a58)
          ![Image](https://github.com/user-attachments/assets/9bf609db-d5dd-4b22-857e-57cf694c7333)

结果包含了对训练集中两张图片的梯度预测和梯度gt,我发现对图像梯度的学习效果并不理想

并且我在训练时也发现梯度学习的loss早期之后就几乎没有下降

以上的实验均是在matting的设置下进行的

因此,我有点怀疑外部监督(梯度的监督)的作用是否足够,我的感受是挺有限的

@ZhengPeng7
Copy link
Owner

我其实也感觉比较有限😂, 特别是在大规模的使用场景训练下. 说实话这种辅助task的影响是不会很直接的, 可以后续加重它的权重看看.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants