2
$\begingroup$

I am working on a project that involves recognising emotion from images. There are two parts in the project: One where I will generate features from images using API's and then classify the emotion and the other part is where I will first use a CNN on a dataset and then take some layer of features and then give it to SVM to classify.

So, I need a dataset of images as well as pre-trained CNN on that dataset also. I have searched quite a lot on google and I am not able to find any. In some cases, I found a pre-trained CNN dataset but could not find the original dataset of images on which CNN was trained.

$\endgroup$

1 Answer 1

2
$\begingroup$

You are probably interested in IAPS, and want to read Analysis of Physiological Signals for Emotion Recognition Based on Support Vector Machine.

But to be honest this classification is doomed in advance, not only pictures are stripped of below threshold of effector activisation but also you do not get baseline as point of reference.

If you are really interested then start from reading everything that Ekman or Sejnowski wrote and feed your NN with only relevant parts, otherwise the most of the data feeded is noise.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.