-
Notifications
You must be signed in to change notification settings - Fork 4
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
Results. #10
Comments
I will do it today. It is slow to upload the results to Google Drive. |
你好,目前也没有发布 AdaRevD 的权重,只有提供了RevD的权重。并且没有详细的带 Adaptive Classifier 的 AdaRevD 的训练代码。 |
AdaRevD 的权重是RevD的权重+classifier的权重(二者都已发布)。Adaptive Classifier的训练方式在readme中已有介绍。 |
通过查看val.py的执行流程确实看到了如何将图像分块之后经过Adaptive Classifier预测模糊程度,从而从不同decoder输出。但是在训练的代码中,我并没有发现Adaptive Classifier预测结果影响输出。
…________________________________
发件人: XIntian Mao ***@***.***>
发送时间: 2024年7月22日 15:54
收件人: INVOKERer/AdaRevD ***@***.***>
抄送: deltawen ***@***.***>; Comment ***@***.***>
主题: Re: [INVOKERer/AdaRevD] Results. (Issue #10)
AdaRevD 的权重是RevD的权重+classifier的权重(二者都已发布)。Adaptive Classifier的训练方式在readme中已有介绍。
―
Reply to this email directly, view it on GitHub<#10 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AT6OE5ILMTUWNWLQ73ANROLZNS3B3AVCNFSM6AAAAABLHGK2LOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBSGMYTOOJQHA>.
You are receiving this because you commented.Message ID: ***@***.***>
|
Classifier训练只会训练Classifier用于预测图像的降质程度,不会对结果产生影响。 |
Pretrain过程中也没有利用Classifier的结果,直接使用最后一个deocder的输出结果来计算loss。除了Classifier和Pretrain还有其他过程吗?
所以Classifier的分类级别(不同PSNR间隔)和decoder之间是如何关联起来的。
…________________________________
发件人: XIntian Mao ***@***.***>
发送时间: 2024年7月22日 17:55
收件人: INVOKERer/AdaRevD ***@***.***>
抄送: deltawen ***@***.***>; Comment ***@***.***>
主题: Re: [INVOKERer/AdaRevD] Results. (Issue #10)
Classifier训练只会训练Classifier用于预测图像的降质程度,不会对结果产生影响。
―
Reply to this email directly, view it on GitHub<#10 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AT6OE5KMPYBQ2WMEHAF3PGDZNTJJHAVCNFSM6AAAAABLHGK2LOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBSGU2TOOBSHA>.
You are receiving this because you commented.Message ID: ***@***.***>
|
Pretrain过程只训练sub-decoder,classifier的分类级别只用于inference时候的adaptive exit。具体内容在paper中有说明。 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Can u also upload HIDE dataset Results for Motion Deblurring trained on GoPro Dataset. on Google Drive
The text was updated successfully, but these errors were encountered: