Hi,
I have read your paper and I have some confusion on 2.1.3 CNN Architecture.

Therefore, for our best CNN structure, we trained on all FER+TFD and performed early stopping on AFEW-faces train+validation.

As we all know, the AFEW-faces are clipped in the video, and the video labels are targeted to individual frames. So I think it is not good to perform early stopping on AFEW-faces, because the the labels of AFEW-faces have too much noise. Why didn't you split the FER+TFD as training and validation, and perform early stopping on the validation.

And another question is that how you decide the early stopping? I am trying to use no-improvement-in-five epoch rule .
Thank you! Look forward to your reply!

Now,I'm studing your paper Recurrent Neural Networks for Emotion Recognition in Video .And I'M doing the expreiment accoroding with the code in the github.But I can't find the
'/home/vincent/data/emotiw2015/Datasets/convnet_data.h5''/home/vincent/data/emotiw2015/Datasets/fertfd_mean_std.h5'

In your RNN folder, the

rnn_on_convnet_features_multi_frame.py,

https://github.com/saebrahimi/Emotion-Recognition-RNN/blob/master/RNN/rnn_on_convnet_features_multi_frame.py#L14
your code is from rnn_multi_frame import RNN, but I can not find the rnn_multi_frame module,
I think it should be from rnn import RNN

Hi,
I am trying to run your code using all the data sets that you have been used except TFD data. I have done pre processing following your code,but while working with CNN codes,train_emoti_convnet_flip.py gives me error. I have attached my error file here. Can you please give me solution.
error.txt

Thank you!

Hi,
I am very interested in your code, but it is hard to understand your code. Could you present a demo for your RNN code for audio or video emotion recognition?

Hi,
In your PreProcessing folder, in the mapAll2FER.m file, https://github.com/saebrahimi/Emotion-Recognition-RNN/blob/master/PreProcessing/mapAll2FER.m#L18
you comment

%add border of 32 for TFD which results in 160 image size

but actually you add 48 to the pavg

@saebrahimi
I have read your paper and interested by your net work for video analysis. I am a fresh wanting to build net work for video while it seems hard for me to understand you code. Would you please offer an instruction? Thank you very much.