I created this smile/neutral CNN while I was bored during COVID-19 isolation to practice my ML skills. NOTE: It can only detect neutral and smile faces, not intended to detect frowning, laughing, etc...
- Install Python.
- Run
python3 -m pip install -r requirements.txt
inside thesmile-detection/
directory.
NOTE: By default, this GitHub repository comes with a pre-trained model. Training a new model will take a while to run. Please be patient.
- Open
smile-detection/Train.ipynb
with Jupyter Notebook. - In Jupyter Notebook, click:
Cell -> Run All
. - The model data will be written inside the
smile-detection/
directory assmile.model/
.
- Run
python3 main.py
inside thesmile-detection/
directory.
The data can be found inside smile-detection/data.csv
, where each image is represented as an Numpy array for simplicty.
One can iterate through each image and read the numpy array as an image.