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

Would you mind offering the trained weight of F3Net on low quality? #9

Open
Telephone1024 opened this issue Sep 17, 2021 · 2 comments
Open

Would you mind offering the trained weight of F3Net on low quality? #9

Telephone1024 opened this issue Sep 17, 2021 · 2 comments

Comments

@Telephone1024
Copy link

你好，我在我划分的ff++上运行你的代码，一段视频采帧60帧左右，但是在17k个iter之后，Both模型的auc还是没有明显提升，卡死在0.75附近，请问这个和数据集规模有关吗，如果方便的话，可否提供你训练的相关配置以及最终保存的模型权重呢
@Telephone1024 Telephone1024 changed the title Would you minding offering the trained weight of F3Net on low quality? Would you mind offering the trained weight of F3Net on low quality? Sep 17, 2021
@Telephone1024 Telephone1024 reopened this Sep 17, 2021
@yyk-wew
Copy link
Owner

yyk-wew commented Sep 17, 2021

您好，训练的相关配置就如代码和readme中所提到的，由于时间比较久了我也没有模型权重可以提供了，抱歉。
根据我之前跑实验的情况，不同数据集setting下网络的性能差异还蛮大的。建议您先在预处理好的数据集上跑一遍backbone(Xception)看看效果，把Xception调到较好水平再在这套参数基础上调其他Module。

@Telephone1024
Copy link
Author

您好，训练的相关配置就如代码和readme中所提到的，由于时间比较久了我也没有模型权重可以提供了，抱歉。
根据我之前跑实验的情况，不同数据集setting下网络的性能差异还蛮大的。建议您先在预处理好的数据集上跑一遍backbone(Xception)看看效果，把Xception调到较好水平再在这套参数基础上调其他Module。

谢谢你的答复，我训练xception在ff++的LQ上可以得到90.3的AUC，但是其acc只有81.6，我个人认为是较好的水平。我注意到，在trainer.py中，你并没有设计lr_scheduler，不知道训练效果不佳是否与此有关。

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
@yyk-wew @Telephone1024