EmotionDetection

High school project on machine learning using TensorFlow, Keras and OpenCV

A model that recognizes emotions and classifies them into 7 categories.

Dataset used (emotion detection) : Fer13
Final model accuracy (emotion detection) : 74.5% on training set, 64% on test set

Dataset used (age detection) : UTKFace
Final model mae (age detection) : 5.6

Download the models here:

Age Detection model : https://drive.google.com/file/d/12WBK04uGUh4yYU8WzYbY7ZNapQ6gkGcv/view?usp=share_link
Emotion Detection model : https://drive.google.com/file/d/1M9SC4r8e3STFP6CV2lA_FLtEpD9B3W51/view?usp=drive_link

Setup:

• Clone the repository
• Download both models and place them in the 'EmotionDetection/' directory
• Make sure dependencies are installed

Usage:

To run using streamlit:

• streamlit run app.py

To run using terminal:

• python detector.py -w (for windowed detection mode)
• python detector.py -c (for camera detection mode)

Dependencies:

• TensorFlow, Opencv, Pillow
• streamlit (optional)