Cwuwhu / FCD-GAN-pytorch

A pytorch project of the algorithm FCD-GAN. Fully Convolutional Change Detection Framework with Generative Adversarial Network (FCD-GAN) is a newly proposed framework for change detection in multi-temporal remote sensing images.
GNU General Public License v3.0
68 stars 7 forks source link

数据预测 #7

Closed jarvii218 closed 1 year ago

jarvii218 commented 1 year ago

您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗?

Cwuwhu commented 1 year ago

其实WSSS部分最下面有测试脚本,自己写也可以,很简单

@.***

发件人: jarvii218 发送时间: 2023-08-28 16:50 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Subscribed 主题: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

jarvii218 commented 1 year ago

是的,已经看到了测试部分,感谢回复

发自我的 iPhone

在 2023年8月29日,11:07,Cwuwhu @.***> 写道:



其实WSSS部分最下面有测试脚本,自己写也可以,很简单

@.***

发件人: jarvii218 发送时间: 2023-08-28 16:50 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Subscribed 主题: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

— Reply to this email directly, view it on GitHubhttps://github.com/Cwuwhu/FCD-GAN-pytorch/issues/7#issuecomment-1696701048, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARCQKDNHVD6SP5JWW3KHMJDXXVMHLANCNFSM6AAAAAA4BCUD74. You are receiving this because you modified the open/close state.Message ID: @.***>

jarvii218 commented 1 year ago

想再请教下,USSS无监督部分训练的时候也需要输入label图像吗?不输入后面S和Dmodel训练似乎会有问题,这和我理解的无监督不太一样?

在 2023年8月29日,11:36,贾 玮 @.***> 写道:

 是的,已经看到了测试部分,感谢回复

发自我的 iPhone

在 2023年8月29日,11:07,Cwuwhu @.***> 写道:



其实WSSS部分最下面有测试脚本,自己写也可以,很简单

@.***

发件人: jarvii218 发送时间: 2023-08-28 16:50 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Subscribed 主题: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

— Reply to this email directly, view it on GitHubhttps://github.com/Cwuwhu/FCD-GAN-pytorch/issues/7#issuecomment-1696701048, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARCQKDNHVD6SP5JWW3KHMJDXXVMHLANCNFSM6AAAAAA4BCUD74. You are receiving this because you modified the open/close state.Message ID: @.***>

Cwuwhu commented 1 year ago

输入的label图像是为了做精度评价,如果你没有label图像随便生成一个图像就行。这个代码本意是用来做研究的,没做的那么灵活

@.***

发件人: JohnnyWick 发送时间: 2023-08-31 09:38 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Cwuwhu; Comment 主题: Re: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 想再请教下,USSS无监督部分训练的时候也需要输入label图像吗?不输入后面S和Dmodel训练似乎会有问题,这和我理解的无监督不太一样?

在 2023年8月29日,11:36,贾 玮 @.***> 写道:

 是的,已经看到了测试部分,感谢回复

发自我的 iPhone

在 2023年8月29日,11:07,Cwuwhu @.***> 写道:



其实WSSS部分最下面有测试脚本,自己写也可以,很简单

@.***

发件人: jarvii218 发送时间: 2023-08-28 16:50 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Subscribed 主题: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

— Reply to this email directly, view it on GitHubhttps://github.com/Cwuwhu/FCD-GAN-pytorch/issues/7#issuecomment-1696701048, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARCQKDNHVD6SP5JWW3KHMJDXXVMHLANCNFSM6AAAAAA4BCUD74. You are receiving this because you modified the open/close state.Message ID: @.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.>

jarvii218 commented 1 year ago

好的,因为我尝试不输入label图像了,也训练了三个模型,但是看见输出路程里在计算精度和loss值,如果不输入label这些值肯定也是空的,我以为需要根据label来评估这些参数后再进一步迭代。不需要的话就是我理解的无监督了,非常感谢您的回答!

发自我的 iPhone

在 2023年8月31日,17:07,Cwuwhu @.***> 写道:



输入的label图像是为了做精度评价,如果你没有label图像随便生成一个图像就行。这个代码本意是用来做研究的,没做的那么灵活

@.***

发件人: JohnnyWick 发送时间: 2023-08-31 09:38 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Cwuwhu; Comment 主题: Re: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 想再请教下,USSS无监督部分训练的时候也需要输入label图像吗?不输入后面S和Dmodel训练似乎会有问题,这和我理解的无监督不太一样?

在 2023年8月29日,11:36,贾 玮 @.***> 写道:

 是的,已经看到了测试部分,感谢回复

发自我的 iPhone

在 2023年8月29日,11:07,Cwuwhu @.***> 写道:



其实WSSS部分最下面有测试脚本,自己写也可以,很简单

@.***

发件人: jarvii218 发送时间: 2023-08-28 16:50 收件人: Cwuwhu/FCD-GAN-pytorch 抄送: Subscribed 主题: [Cwuwhu/FCD-GAN-pytorch] 数据预测 (Issue #7) 您好,请问WSSS部分是否只有训练模型的部分,如果我想进行预测自己的数据集,是需要自己编写测试脚本吗? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

— Reply to this email directly, view it on GitHubhttps://github.com/Cwuwhu/FCD-GAN-pytorch/issues/7#issuecomment-1696701048, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARCQKDNHVD6SP5JWW3KHMJDXXVMHLANCNFSM6AAAAAA4BCUD74. You are receiving this because you modified the open/close state.Message ID: @.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.>

— Reply to this email directly, view it on GitHubhttps://github.com/Cwuwhu/FCD-GAN-pytorch/issues/7#issuecomment-1700653093, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARCQKDJ7MKA2XXBDHPDPHALXYBH3NANCNFSM6AAAAAA4BCUD74. You are receiving this because you modified the open/close state.Message ID: @.***>