FaceAgeMood is a dual deep learning project that uses facial images to perform age prediction and emotion detection. This project demonstrates how a single dataset can be used for multiple facial analysis tasks.
This project uses the UTKFace (new) dataset from Kaggle.
The dataset contains over 20,000 facial images labeled with:
- Age
- Gender
- Ethnicity
Although the dataset doesn't include emotion labels by default, I leveraged facial expressions from the images for an experimental emotion detection task.
| Notebook | Description |
|---|---|
Age_prediction.ipynb |
Predicts age from facial images using a convolutional neural network (CNN). |
emotion-detection.ipynb |
Detects emotion from the same UTKFace images using a custom approach. |
- Image preprocessing using OpenCV and TensorFlow/Keras
- CNN architectures for both age and emotion tasks
- Custom data labeling and augmentation for emotion prediction
- Evaluation metrics and visualizations for performance insight
Main libraries used:
TensorFlow / Keras
OpenCV
Matplotlib
NumPy
scikit-learn
π License
This project is open-sourced under the MIT License. πββοΈ Author
Made with β€οΈ by Tareq