Emoticon 🙎➡️😎

Transfer yourself into live emoji with power of machine learning.

Index

• Install
• Getting started
  • Customization
  • Usage
• Details
• Limitations
• Todo

Install

1. Clone the repo and cd into it.

• (Optional)(For Developers)

  • Create a virtual environment and then go to step 2.

  • Run the following command to create a virtual environment.

    • For Windows:-

    python -m venv env
    

    • For Linux/MacOS:-

    python3 -m venv env
    

  • Run the following command to activate the virtual environment.

    source env/bin/activate
    

2. Install the dependencies using pip.

pip install -r requirements.txt

3. Run index.py file.

   • For Windows:-

   python index.py
   

   • For Linux/MacOS:-

   python3 index.py
   

Getting Started

Customization

• You can use the any camera attached to your system by changing the index of VideoCapture function. If you have 2 cameras then 0 will represent the default camera and other camera will have another index.(sometimes it might be not linear, sometime even for 2 camera setup, second camera's index might be 2 or 3, specially you are using virtual camera)

• Giving the VideoCapture function path to a video will run the program on that video. This is done mostly for testing purposes.

• Change the emoji as per your need in emojis folder. DON'T change the name. Make sure to rename your emoji photos as already given.

Usage

• Create sequence of different mood showing emoji.

• Use the emoticon screen as a webcam for your meetings if you just want to give reactions in meeting.(That's what 90% people do in online meetings)

Details

• This uses DeepFace library to predict the emotion. You can see more details here.

• It has quite good prediction rate. 80%-90% time it predicts the right emotion unless you are not good at expressing them.

Limitations

There are some limitations as of now. Might get eliminated in future if possible.

• Not good performance in low light. You will need quite good quality webcam. One solution is to use your phone's camera as a webcam.

• Does not interact with side faces. It needs straight looking face.

• Emotion flatuates very much sometimes. It can go from sad to happy in one frame which is not practical.

Todo

• Fix the flatuation problem with counter and average emotion by keeping track of past emotions.

• 🏃‍♂️Implement different versions of emoji for different genders. Can be followed up by other parameters like eye color, skin color etc.

• (Should be done parallely) Transfer the emojis to SVG and if possible then animatable SVGs for better customization and cool effects. If not then GIFs are also an reliable option.

Contributors

• Het S. Patel - @PhoenixCreation

Give a 🌟 to this repo if you enjoyed being transferred to emojis.