A real time system to detect human emotions from image and voice and predict its reaction. This is a pre-trained model with an accuracy of 80.7% over Facial Expression Recognition (FER) 2013 dataset. This project is the direct outcome of my graduation submission at MIT, Pune. The extraction of emotions from audio is done by using pyAudio Reaction prediction module is under development and is not included yet
Python 3
OpenCV
pip install tensorflow numpy scipy h5py
Download the trained model from the links given in source/trained-model.txt
python em_model.py