FeelTube

A simple app that classifies a user’s emotion based on facial landmarks!

Running the project

• The project as of now has 3 running functionalities

• To run the Emotion Classifier on Jeffa dataset:

cd /path/to/your/directory
python emotrainer.py

• After which the program will show you a set of 5 images that are chosen at random as test images:
  

• Closing that shows us the histogram plotted using LPB:
  

• Closing that gives us the output with red text representing wrong prediction and green being right:
  

• The terminal has information about the mean accuracy:
  

• Sometimes output will have errors :
  

• To run the webscraping application for emotion to genre selection:

cd /path/to/your/directory
python emovie.py

• To run the face and features detector:
  This uses the 68 landmarks method:
  

cd /path/to/your/directory
python facial_landmarks.py --shape-predictor shape_predictor_68_face_landmarks.dat --image images/<imagename>.jpg