Open Kexiii opened 6 years ago
I think you should be able to use any of the PyTorch pretrained DenseNets:
model_urls = {
'densenet121': 'https://download.pytorch.org/models/densenet121-a639ec97.pth',
'densenet169': 'https://download.pytorch.org/models/densenet169-b2777c0a.pth',
'densenet201': 'https://download.pytorch.org/models/densenet201-c1103571.pth',
'densenet161': 'https://download.pytorch.org/models/densenet161-8d451a50.pth',
}
Those urls should have the weights. Let me know if this works!
Thanks for your reply! I forgot to mention that I'm using the old pytorch0.3 implementation and it can't directly use the pretrained weights you provided.
Did you try the ones here: https://github.com/mingminzhen/densenet-efficient-model
Yes, but it is based on the pretrained memory efficient torch model(232,264): https://github.com/liuzhuang13/DenseNet In the tech report, they didn't provide efficient densenet169 pretrained torch model that I want
Try converting the pytorch model then. All you have to do is change the name of the keys in the state duct in the link that I supplied. That’s how we created the other efficient densenets.
Best, Geoff
On May 23, 2018, 10:55 PM -0400, Kexiii notifications@github.com, wrote:
Yes, but it is based on the pretrained memory efficient torch model(232,264): https://github.com/liuzhuang13/DenseNet In the tech report, they didn't provide efficient densenet169 pretrained torch model that I want — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.
Thanks for your advice, want to let you know I've tried to convert the densenet169:https://github.com/Kexiii/Pretrained-Efficient-DenseNet
I think you should be able to use any of the PyTorch pretrained DenseNets:
model_urls = { 'densenet121': 'https://download.pytorch.org/models/densenet121-a639ec97.pth', 'densenet169': 'https://download.pytorch.org/models/densenet169-b2777c0a.pth', 'densenet201': 'https://download.pytorch.org/models/densenet201-c1103571.pth', 'densenet161': 'https://download.pytorch.org/models/densenet161-8d451a50.pth', }
Those urls should have the weights. Let me know if this works!
Hi, I have a question On which dataset these weights are pretrained,ImageNet?
Yup. Imagenet.
Hi, thanks for your great work! I'm working on densenet169 these days, do you know where I can find the ImageNet pretrained weights for this efficient implementation? Or do you have any example code to show how to convert the other implementation's pretrained model to this one? I do have noticed this #13 , but it seems @ZhengRui didn't provide any example code, and I don't know where to start..