Pretrained Models (on ImageNet) EfficientNet-v2 link not good

[lannguyen0910]

Pretrained Models (on ImageNet) EfficientNet-v2 link not good

[lannguyen0910]

Hi @mbenami, thanks for asking the question. You could try the script below to download the Efficientnet-v2 ImageNet weight

!gdown https://drive.google.com/uc?id=1rTYuStYA64U1qb7uC-h8DOhv99Ztxaxz

And if you want to load it, you should change 2 lines in either "efficientnetv2_market.yaml" or "efficientnetv2_cuhk.yaml" to:

pretrain_path: "efficienetv2_imagenet.pt"
pretrain_choice: "imagenet"

I suggest that you should follow the instructions in the notebook i attached in README. If you come up with any errors or issues, please reach out to me soon. Thank you!

[mbenami]

Hi @lannguyen0910
thanks for the quick response
still when I try to load the weight I get an error

load weights

torch.load("efficienetv2_imagenet.pt")

error:

ModuleNotFoundError: No module named 'nets'

you can also see there is not weights in the file :
https://drive.google.com/file/d/1rTYuStYA64U1qb7uC-h8DOhv99Ztxaxz/view

[lannguyen0910]

Hi @mbenami, have you tried changing 2 lines in the yaml file and run it? If you follow the instructions in the colab notebook, you shouldn't get any errors in loading the weight, although some gdown lines are inaccessible now but you can still download the market1501 dataset and efficienetv2_imagenet.pt weight.

If you didn't run the notebook, can i ask where is the specific file you placetorch.load("efficienetv2_imagenet.pt") line? Because the code itself doesn't have that line, you can take a look at line 153 and 201 in nets/nn.py for more information.

Besides, when i click in the weight link, i can still download the weight, here are some images: