A small application to classify the mood of one or more faces captured by a webcam.
To run this application locally you need to have the following things installed.
- Python 3 >
- OpenCV
- Dlib
- Sklearn
- Pickel
The dataset I used is Cohn-Kanade (CK and CK+) database. http://www.consortium.ri.cmu.edu/ckagree/
Once the dataset is downloaded run python3 prep.py to cleanup the dataset and sort it correctly.
After sorting and cleaning up the model needs to be trained. This can be done by running python3 trainer.py.
And to use the mood detection after creating the model run python3 app.py.
- The emotions that should be detected are now defined in 3 sperate files, this should be unified.
- The accuracy can be improved by using more data and making sure that each data class has a similar amount of training data.