Facial Expression Recognition

a convolutional neural network for recognizing facial expressions on live webcam images.

Installation

The implementation has been tested with Python 3.6.3. You can create a fresh virtual environment with conda or virtualenv if you like. TensorFlow is not supporting conda officially, so pip is used for package management. All the dependencies can be found in the requirements.txt file.

With your Python 3 environment activated, you can install the requirements with

pip install -r path/to/requirements.txt

Live prediction

If your computer has a webcam, you can compute predictions on the fly. Take off your glasses and hats and start the live-prediction with

python webcam.py

Training

If you would like to train the tensorflow CNN yourself, you need to obtain the FER2013 dataset from kaggle and, optionally, the CK+ dataset. For CK+, you can use the ckplus_to_csv.py script to automatically detect all the faces, parse the grayscale intensities and collect all the CK-images into a single CSV file.

The directory structure should look like this:

project root
│   README.md
│   ...   
└───data
│   │   fer2013.csv
│   │   ckplus.csv
│   │
│   └───ckplus
│       │   images
│       │   labels

You can invoke the (somewhat lengthy) training process with

python train_fer.py