Challenge-condition-FER-dataset
Challenge-condition-FER-dataset copied to clipboard
RAN VGG16 训练模型
作者您好!关注了您在TIP和CVPR上发表的论文RAN与SCN,现在准备根据您提供的github 开源代码复现RAN,据我所知,在RAN论文中提到在FERPlus上最高的点数是89.16,而且是以VGG16为backbone实现的,但是在开源的代码中没有基于VGG16的网络实现,请问您是否可以开源这一块的代码?此外,我想请问在FERPlus,您是否有尝试resnet50、VGG19等网络层次更深的网络作为backbone?它们有提高FER的准确度吗?期待您的回复,谢谢!
VGG16是从vgg官方公布的vggface的caffe模型转成pytorch去训的,记得要用soft label哈。我试过nus 赵健博士公布的IR-50系列代码,是可以进一步提高的。具体请参考赵健博士的:https://github.com/ZhaoJ9014/face.evoLVe.PyTorch。
好的,谢谢您的回复!期待您尽快开源SCN!
您好!我阅读了您的test_rank_loss_attention.py,发现在143行,送入model的input_var并不是原图加上5张crop后的图像,这与train_rank_loss_attention.py输入model的不符呀,根据论文也是原图加上5张crop图像作为RAN网络的输入?checkpoint_[1, 1]_33_0.002_89.162.pth.tar是VGG16上的训练模型?resnet18的训练模型能够开源吗?此外在attention_rank_loss.py中alphas_org对应的原图是最后输入的吗?论文中指出的原图是在0的位置,而代码中应该是在5的位置,我不知道我的理解是否正确?期待您的答疑!谢谢!
你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org. jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机
好的,感谢您的回复,我再好好看看代码
你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机
Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb)
这里是为了检查模型的keys和网络定义的keys一致与否,不一致的话load不进来参数,会随机初始化。所以记得看下是否一样哈,一样的话 可以放心大胆load
LiuYeeJ @.***> 於 2021年5月10日週一 下午7:44寫道:
你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机
Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb)
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836594505, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4DF7OLEBUK4YN3OCRTTM7BJFANCNFSM4MDSR65Q .
这里是为了检查模型的keys和网络定义的keys一致与否,不一致的话load不进来参数,会随机初始化。所以记得看下是否一样哈,一样的话 可以放心大胆load LiuYeeJ @.***> 於 2021年5月10日週一 下午7:44寫道: … 你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机 Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4DF7OLEBUK4YN3OCRTTM7BJFANCNFSM4MDSR65Q .
我不知道理解的对不对昂,但是输入输出是完全一致的,说明两个key是完全一致的,但是程序就是停在这里了,并不能继续加载,可能问题比较愚蠢哈,感谢回复,万分感谢!!!!!
那你把断点去掉看看?
LiuYeeJ @.***> 於 2021年5月10日週一 下午8:55寫道:
这里是为了检查模型的keys和网络定义的keys一致与否,不一致的话load不进来参数,会随机初始化。所以记得看下是否一样哈,一样的话 可以放心大胆load LiuYeeJ @.***> 於 2021年5月10日週一 下午7:44寫道: … <#m_-1319756849151180371_> 你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机 Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 (comment) https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836594505>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4DF7OLEBUK4YN3OCRTTM7BJFANCNFSM4MDSR65Q .
我不知道理解的对不对昂,但是输入输出是完全一致的,说明两个key是完全一致的,但是程序就是停在这里了,并不能继续加载,可能问题比较愚蠢哈,感谢回复,万分感谢!!!!!
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836664820, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4BACMKXF3OSAZWSR2TTM7JVJANCNFSM4MDSR65Q .
那你把断点去掉看看? LiuYeeJ @.> 於 2021年5月10日週一 下午8:55寫道: … 这里是为了检查模型的keys和网络定义的keys一致与否,不一致的话load不进来参数,会随机初始化。所以记得看下是否一样哈,一样的话 可以放心大胆load LiuYeeJ @.> 於 2021年5月10日週一 下午7:44寫道: … <#m_-1319756849151180371_> 你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机 Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 (comment) <#18 (comment)>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4DF7OLEBUK4YN3OCRTTM7BJFANCNFSM4MDSR65Q . 我不知道理解的对不对昂,但是输入输出是完全一致的,说明两个key是完全一致的,但是程序就是停在这里了,并不能继续加载,可能问题比较愚蠢哈,感谢回复,万分感谢!!!!! — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4BACMKXF3OSAZWSR2TTM7JVJANCNFSM4MDSR65Q .
感谢老哥!!!!我简直就是ZZ,服了,太感谢了,居然一直回复我,万分感谢
不用客气,记得close issue
LiuYeeJ @.***> 於 2021年5月10日週一 下午10:23寫道:
那你把断点去掉看看? LiuYeeJ @.
> 於 2021年5月10日週一 下午8:55寫道: … <#m_-4045927442501514654_> 这里是为了检查模型的keys和网络定义的keys一致与否,不一致的话load不进来参数,会随机初始化。所以记得看下是否一样哈,一样的话 可以放心大胆load LiuYeeJ @.> 於 2021年5月10日週一 下午7:44寫道: … <#m_-1319756849151180371_> 你先好好看看code哈,关于org顺序的问题,我当时是sort的,也就是图片每个文件夹下的名字都是一样的,org。jpg排在第五很正常吧。论文里画图是为了方便理解发自我的华为手机 Hello, I would also like to ask, when I run the training model code, finally need me to enter key, because the program is not very understanding, so I did not read this input is what do I need to enter? Hope to be able to get your reply, if there is any interruption please forgive. train_attention_rank_loss.py(165) -> for key in pretrained_state_dict: (Pdb) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18 (comment) <#18 (comment) https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836594505>>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4DF7OLEBUK4YN3OCRTTM7BJFANCNFSM4MDSR65Q . 我不知道理解的对不对昂,但是输入输出是完全一致的,说明两个key是完全一致的,但是程序就是停在这里了,并不能继续加载,可能问题比较愚蠢哈,感谢回复,万分感谢!!!!! — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#18 (comment) https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836664820>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4BACMKXF3OSAZWSR2TTM7JVJANCNFSM4MDSR65Q .
感谢老哥!!!!我简直就是ZZ,服了,太感谢了,居然一直回复我,万分感谢
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kaiwang960112/Challenge-condition-FER-dataset/issues/18#issuecomment-836763258, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6LK4HKOSYA7OZOZUN2DHDTM7T6TANCNFSM4MDSR65Q .
好的,因为我不是问题发起者没办法关闭,但是我把自己的提问隐藏了,感谢作者!