Facial Expression Detection Using Keras And Deploy With Flask

In this project, I made a custom CNN-model which can detect the 7 universal facial expressions and deployed the model using Flask.

The 7 universal facial expression includes

• Surprise
• Happy
• Neutral
• Angry
• Sad
• Disgust
• Fear

Sample Output

Dataset

I used the dataset from this Kaggle competition.

Model Architecture

This is the model architecture that I used for facial expression detection.

Training

Got 67% validation accuracy within 15 epoch which took 7 min 30 sec.

For more details check Facial_Expression_Training.ipynb.

Deploying using Flask

To test the deployed model

• Keep the input video in videos folder.

• run main.py.

Python3 main.py