adibyte95 / Emotion-detection

detection of human emotions

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

HitCount

Emotion detection

To detect seven human emotions
1.Angry
2.Disgust
3.Fear
4.Happy
5.Sad
6.Surprise
7.Neutral


Dataset

Kaggle dataset was used here

Accuracy

Training accuracy : 66 percent
Testing accuracy : 62 percent

Confusion matrix

confusion matrix for the test set containing nearly 7k examples
confusion matrix

learning curve

learning curve
learning curve

Note

After 40 epochs the model was starting to overfit with very minute increase in test accuracy so i used early stopping and training and testing accuracy reported here is uptill here. Dropout is used to reduce the overfitting for the model. However an accuracy of ~63-64 percent can be achieved if the model is trained further.
I am open to pull requests and suggestions for improvement of results

About

detection of human emotions

License:MIT License


Languages

Language:Jupyter Notebook 100.0%