Facial Emotion Detection

A robust web service for real-time facial emotion recognition. This project leverages a Convolutional Neural Network (CNN) to classify facial expressions from uploaded images and stores the results in a PostgreSQL database.

---

Features

• Face Detection: Utilizes OpenCV's Haar Cascade to accurately locate faces in images.
• Emotion Recognition: Classifies faces into 7 categories: Angry, Disgust, Fear, Happy, Neutral, Sad, and Surprised.
• API Endpoints:
  • POST /predict_emotion: Upload an image (JPEG/PNG) to get an instant emotion prediction.
  • GET /predictions: Retrieve a history of all analyzed images and their predicted emotions.
• Database Integration: Automatically stores prediction results, confidence scores, and timestamps using SQLAlchemy.

---

Tech Stack

• Framework: FastAPI
• Deep Learning: TensorFlow, Keras, OpenCV
• Database: PostgreSQL (SQLAlchemy ORM)
• Data Processing: NumPy, Pandas, Matplotlib

---

Project Structure

.
├── DL/                     # Deep Learning components
│   ├── detect_and_predict.py # Emotion detection logic
│   ├── emotion_cnn_model.keras # Pre-trained CNN model
│   ├── haarcascade.xml      # Haar Cascade for face detection
│   └── emotion_cnn_model.ipynb # Training notebook
├── models/                 # Database & API schemas
│   └── model.py            # SQLAlchemy & Pydantic models
├── config.py               # Environment configuration
├── database.py             # Database connection setup
├── main.py                 # FastAPI application entry point
├── requirements.txt        # Project dependencies
└── README.md               # Project documentation

---

Installation & Setup

1. Clone the repository

git clone https://github.com/elhidarinouhayla/Facial-Emotion-Detection.git
cd Facial-Emotion-Detection

2. Configure Environment Variables

Create a .env file in the root directory with your PostgreSQL credentials:

USER=your_db_user
PASSWORD=your_db_password
HOST=localhost
PORT=5432
DATABASE=facial_emotion_db

3. Install dependencies

pip install -r requirements.txt

4. Run the Application

uvicorn main:app --reload

The API will be available at http://localhost:8000. You can access the interactive documentation at http://localhost:8000/docs.

---

API Usage

Predict Emotion

• Endpoint: /predict_emotion
• Method: POST
• Payload: Form-data with key file (Image)
• Response:

{
  "id": 1,
  "emotion": "happy",
  "confidence": 98.45,
  "created_at": "2024-04-08T12:00:00Z"
}

Get All Predictions

• Endpoint: /predictions
• Method: GET
• Response: List of prediction records.

---

Testing

To run the tests:

pytest

---