hx173149 / C3D-tensorflow

C3D is a modified version of BVLC tensorflow to support 3D ConvNets.
MIT License
588 stars 262 forks source link

Loading pretrained model #18

Closed gandhirushabh2311 closed 7 years ago

gandhirushabh2311 commented 7 years ago

I wanted to load the pretrained model which is in the form of a .model file. However as far as I know saved models in tensorflow are in the form of .meta files and can be loaded by tf.train.save_meta_files(). Can someone please let me know how can I load the graph from this model?

geroge-gao commented 6 years ago

@gandhirushabh2311 have you ever solved the problem?

JulioEndara commented 6 years ago

@geroge-gao download the .model files as per the instructions in readme file.

geroge-gao commented 6 years ago

@JulioEndara I downlaoded all the .model files, but when I use the pre-trained model,there occurred an error. ValueError: The specified path: conv3d_deepnetA_sport1m_iter_1900000_TF.model is a file. Please specify only the path prefix to the checkpoint files.

do you know know how to solve the problem?

JulioEndara commented 6 years ago

@geroge-gao no I didn't have that issue.

geroge-gao commented 6 years ago

@JulioEndara could you tell me how you use the pre-trained model file?

JulioEndara commented 6 years ago

@geroge-gao I did follow the instructions as per readme file.

geroge-gao commented 6 years ago

@JulioEndara sorry to brother you,but this problem has troubled me for a few days, could I ask you a few questions on the email? my email is codergeroge@gmail.com

JulioEndara commented 6 years ago

@geroge-gao github is better than email because other people see the issue and they can help out rather than just one person, btw I'm not an expert on this so you are better off raising your issue to the community.

Also this is also a closed issue.

I suggest you open a new issue and explain in detail what you are doing, what script you are executing and include full dump of your error... also confirm if you have modified the code or use it straight from the repo.