Fail to load pth model files

[ReedZyd]

I have encountered a problem：
Traceback (most recent call last):
File "demo.py", line 5, in
dex.eval()
File "/home/zyd/Reed/DL/age-estimation/code/pytorch-DEX/dex/api.py", line 23, in _eval
gender_model.load_state_dict(torch.load(gender_model_path))
File "/home/zyd/anaconda3/envs/pytorch/lib/python3.5/site-packages/torch/serialization.py", line 367, in load
return _load(f, map_location, pickle_module)
File "/home/zyd/anaconda3/envs/pytorch/lib/python3.5/site-packages/torch/serialization.py", line 528, in _load
magic_number = pickle_module.load(f)
_pickle.UnpicklingError: invalid load key, 'v'.

How can I solve it?

[siriusdemon]

You should first install git-lfs, because its pth files are stored in remote lfs server.

[ReedZyd]

The problem still exists and I can't download dex/pth/age_sd.pth(539 MB).

[siriusdemon]

Maybe you can download here and replace the two git-lfs pointer file in dex/dex/pth.
gender_sd.pth
age_sd.pth

[ReedZyd]

I solved this problem by installing git-lfs with apt-get. Also, is there code that tests performance?I want to test the DEX network by MAE.

[ReedZyd]

Maybe you can download here and replace the two git-lfs pointer file in dex/dex/pth.
gender_sd.pth
age_sd.pth

Thank you very much!

[siriusdemon]

I solved this problem by installing git-lfs with apt-get. Also, is there code that tests performance?I want to test the DEX network by MAE.

You can have a try! Refer to its home page for performance metrics.

[siriusdemon]

The Great Firewall ... ... It seems that it is not a good idea to store large file by git-lfs.
Download from Google Drive or Baiduwangpang:
link: https://pan.baidu.com/s/1pnU7ZwzYU4MqCvRQN6xN0Q
code: vdij

[ReedZyd]

Yes,I solve it!

[neverUseThisName]

Maybe you can download here and replace the two git-lfs pointer file in dex/dex/pth.
gender_sd.pth
age_sd.pth

how to modify .pth file after downloading the models?

[siriusdemon]

If any git-lfs pointer exists in dex/dex/pth, remove them.
Once you have downloaded this two files, move them to dex/dex/pth.

[danilocoura]

Maybe you can download here and replace the two git-lfs pointer file in dex/dex/pth.
gender_sd.pth
age_sd.pth

Hi. Do you have the pytorch models only trained with the imdb-wiki datasets? (I think these are also trained with the chalearn database)

Thank you!

[siriusdemon]

No. But there are other DEX implementations on github, Good luck! Happy New Year!

[danilocoura]

Hi, Sirius. Thanks for your answer. Happy New Year! I am stuck in this problem. You released the pytorch model of "dex_chalearn_iccv2015.caffemodel". I tried to use your code (convert.py) to convert the "dex_imdb_wiki.caffemodel" ( https://data.vision.ee.ethz.ch/cvl/rrothe/imdb-wiki/static/dex_imdb_wiki.caffemodel), but I am having problems with the protobuf size. I would like to know how you deal with this or if you could convert this model to me, please. I really appreciate your help. best regards Danilo

…

On Fri, Jan 24, 2020 at 12:16 AM Sirius Demon ***@***.***> wrote: No. But there are other DEX implementations on github, Good luck! Happy New Year! — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#1?email_source=notifications&email_token=AG2ILOQR4KP5RZSJB32TKB3Q7J2SBA5CNFSM4G7NGRF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJZX7PY#issuecomment-577994687>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AG2ILOU7CSB2OINWS6TDKILQ7J2SBANCNFSM4G7NGRFQ> .

[danilocoura]

Hi, Sirius. I've managed my issue. I used the MMDNN to convert the models (caffe to pytorch). So, disregard the last e-mail. Thanks for your attention. best regards, Danilo

…

On Tue, Jan 28, 2020 at 5:18 PM Danilo Coura ***@***.***> wrote: Hi, Sirius. Thanks for your answer. Happy New Year! I am stuck in this problem. You released the pytorch model of "dex_chalearn_iccv2015.caffemodel". I tried to use your code (convert.py) to convert the "dex_imdb_wiki.caffemodel" ( https://data.vision.ee.ethz.ch/cvl/rrothe/imdb-wiki/static/dex_imdb_wiki.caffemodel), but I am having problems with the protobuf size. I would like to know how you deal with this or if you could convert this model to me, please. I really appreciate your help. best regards Danilo On Fri, Jan 24, 2020 at 12:16 AM Sirius Demon ***@***.***> wrote: > No. But there are other DEX implementations on github, Good luck! Happy > New Year! > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#1?email_source=notifications&email_token=AG2ILOQR4KP5RZSJB32TKB3Q7J2SBA5CNFSM4G7NGRF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJZX7PY#issuecomment-577994687>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AG2ILOU7CSB2OINWS6TDKILQ7J2SBANCNFSM4G7NGRFQ> > . >

[quocnhat]

Maybe you can download here and replace the two git-lfs pointer file in dex/dex/pth.
gender_sd.pth
age_sd.pth

solved the problem by either git-lfs installing and replacing 2 pretrained files. thanks