• Download the Fer2013 dataset using https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge/data. Fer2013 contains over 35,000 images grouped in seven emotion category: Anger-0, Disgust-1, Fear-2, Happy-3, Sad-4, Surprise-5 and Neutral-6.
• Download and add Dlib Shape Predictor model using http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
• Download the Haar-casade frontalface Detection XML using https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_frontalface_alt.xml
• The code was developed on Anaconda environment. The following instruction are relaated to it.
• unzip the Fer2013.zip and place the Fer2013.csv file in the code folder
• Place dlib shape predictor and haar-casde frontalface XML in the code folder
• Perform the following operation in Jupiter Notebook workbook to setup the environment

!pip install numpy
!pip install argparse
!pip install sklearn
!pip install scikit-image
!pip install pandas
!pip install hyperopt
!pip install dlib

• Run the image_processing_and_features_extraction.py to perform the image processiong and feature extraction process to generate Fez2013_feature folder containing the traing set and test set

python image_processing_and_features_extraction.py

• Train and test model using emotion_classifier.py

python emotion_classifier.py

Image Processing using face detection and reshaping the images to 48X48 size to uniform the alignment

Feature extraction using

• Gabor Filter
• Histogram of Oriented Gradient (HOG)
• Facial Landmarks Extraction

Support-Vector Machine Model training on training set with RBF kernel function

Validation and testing with generation of accuracy score