pliang279 / mfn Goto Github PK

Hi Paul,
I have a question about the data ( X_train X_test X_valid)
for example, the X_train is (1284,20,325) and is the '325' composed of these three parts ,?
300d - text
20d - visual
5d - audio
and what's the meaning of the '20'?

Hello,

I am trying to run the test_mosi.py code, having installed the right version of each library. However, I keep having errors related to TensorFlow. Can you tell me which version of TensorFlow you use?

Thank you for your help!

Hi,
I found that there are some inconsistencies in the processed MMMO dataset: http://immortal.multicomp.cs.cmu.edu/raw_datasets/processed_data/mmmo/data/ the p files have sequence length 21 and visual feature dimension is 35, but https://github.com/pliang279/MFN/tree/master/new_data/mmmo the h5 files have sequence length 20 and visual feature dimension is 36.
So what happened?
Thanks in advance.

I really appreciate this code. Please use a .gitignore file to exclude pyc files. They should almost never be in a repo. If you search for how to untrack and remove pyc from git, you will find the relevant resources.

Hi,
Thanks for sharing code! But could you please provide more details about how to use our own data to run this model?
Thanks!

You have linked to other papers in the readme, but not to your own. Please always add a link to your Arxiv paper in your readme. The link is https://arxiv.org/abs/1802.00927

Hi Paul,
Thanks for sharing the code.
I have a question about the feature selection, which is not mentioned in your paper.
Since we don't have the file /media/bighdd5/Paul/mosi/fs_mask.pkl, could you tell us which parameters work the best on that dataset and how did you obtained them ?
Cheers,
Valentin

Hi,
I was just wondering if you have released the code for Graph-MFN which has the implementation of DFG with MFN network. If you have, please point me to it as I was unable to find it.

Thanks
Nikunj Kotecha

Thank you very much for you share the DataSet.
Text_train_em.h5 is not in /new_data/pom. Can you upload the file?

Hi,

In 'test_mosi.py', there is no exit criterion for the while loop. When does it stop then?

Thanks,
VR

Hi @pliang279 and @A2Zadeh

I am trying to replicate the results on IEMOCAP dataset regarding arousal and valance, I have downloaded processed data from http://immortal.multicomp.cs.cmu.edu/raw_datasets/processed_data/iemocap/seq_length_20/, but when I open y_valid.p it seems that this data is used for category emotion recognition because it shows one-hot-vector of 11 elements.

Could you help me get the right processed IEMOCAP data to replicate this results and also give some explanations or instructions? really appreciate

Regards
Dung

I understand that this repo was developed in a short amount of time, but Python 2.7 should never be used for any new project. It is unreasonable to require it, given that it is just a year away from expiration. Refer to https://pythonclock.org/ .

Please confirm the code works under the current version of Python, i.e. 3.x.