Real-time Smile Recognition

The goal of this project is to develop a real time smile detection algorithm running on OpenCV using deep learning in keras

Installation

Step 1

Install Python 3.6

Step 2

Navigate to the root directory (where this README is located) and run

pip install -e .

Step 3

Download genki4k dataset. And put it into "project-root/data/" folder.

Running the project

Train the model for the first time from root directory with

python src/train.py

Using the same command again does not train the model again unless the relevant line in the code is changed.

Run the camera script with

python src/run_camera.py

and press q to quit.

Or run one image though the model with

python src/run.py

Change the file path in the code to change the image.

Directory structure

• /
  • requirements.txt: the project requirements
  • setup.py: automatic installation with pip install . -e
• /src/
  • models/: directory for different kinds of neural models
  • utils/: directory for helper functions
  • run_camera.py: running the real-time application for smile detection
  • run.py: run only once on a static file (no evaluation)
  • train.py: train and evaluate the model
• /data/
  • genki4k/: contains the extracted dataset as downloaded from [http://mplab.ucsd.edu/wordpress/?page_id=398]
  • models/: directory for the saved trained models
• /report/: Contains the report files