Cypher 2021 - EmoExpressor

An emoji recommender based on real-time emotion detection.

What it does?

• Capture your face, analyze your emotion, and recommend some emoji and emoticons that suit your mood.
  

Inspiration

• We often use emojis and emoticons to express our feelings when texting or using social media. However, it's not easy to select and paste emojis on computers.
• People often don't know which emoji to choose. AI can decide it for you.
  
• College students face a lot of pressure. So why not have some fun with an AI?
  

How we build it

• Use OpenCV to analyze the real-time video captured from the front camera
• Use Tensorflow and Keras to build the emotion recognition models
• Implement the web-based front end via Flask

Challenges we face

• We are all new to deep learning. It took us effort to understand the logic behind emotion detection.
  
• We are also new to front-end developing. We spent lots of time making the UI responds to the outputs of the deep learning model.
  

Accomplishments that we're proud of

• Achieved a working emotion detection function based on models from atulapra
• Produced a nice web UI with no prior experience
• Learned lots of new techniques within days

What we learned?

• Implementation of CNN in python (not the whole theory, but how to use it)
  
• Using OpenCV package to implement emotion recognition
  
• Developing a web-based front end for a python program with flask
  
• Develop user interface with Tkinter and curses (though they are not used in the final version)
  
• Project management for team work

What's next for EmoExpressor

• A better user interface and a desktop client
• More available emotions and better emotion recognition precision
• Better user experience design, e.g., click to copy, camera switch, etc.

Reference

• Thanks to atulapra for his models
• Data set used in this project: FER-2013

Built With

• flask
• keras
• opencv
• python
• tensorflow