kimbyeolhee/Real_Time_Emotion_Detector

Real Time Emotion Detector

Using cam to capture face emotion into 4 category (Happy, Sad, Neutral, Angry)

Environment setup

We have experimented the implementation on the following enviornment.

# create virtual python enviroment
python -m venv venv_vit
source venv_vit/Scripts/activate

# install required libraries
pip install -r requirements.txt

Prepare dataset

Datasets we used are as follows:

Dataset	Download	Comment
FER-2013	Link	Change categories 7 to 4

For more details, please refer to data description.

Train model

python train.py -c base_config

Detect emotion

Download the haarcascade_frontface_default.xml file and place it in the haarcascade_files folder.

You can use the trained model by specifying the base_config from Hugging Face (StoneSeller/emotion-classifier-vit).

python inference.py -c inference_config