Real-Time Emotion Detection is a state-of-the-art emotion recognition system that analyzes facial expressions in real-time using advanced deep learning models. This project combines multiple AI models (DeepFace, FER2013 Mini-XCEPTION) with MediaPipe face detection to deliver accurate, fast, and robust emotion analysis.

• Multi-Model Ensemble: Combines DeepFace and custom-trained CNN models for superior accuracy
• Real-Time Performance: 60+ FPS with optimized frame processing and multithreading
• Professional UI: Cyberpunk-inspired HUD with live emotion probability bars
• 7 Emotion Classes: Detects Happy, Sad, Angry, Surprise, Fear, Disgust, and Neutral
• Video Recording: Built-in screen recording for content creation
• Modular Architecture: Clean, maintainable code structure following best practices
• Dataset Integration: Direct download from Kaggle FER2013 dataset

• Python 3.8 or higher
• Webcam/Camera
• (Optional) GPU with CUDA for faster processing

1. Clone the repository

   git clone https://github.com/Shayanthn/Real-Time-Emotion-Detection-with-OpenCV-DeepFace.git
   cd Real-Time-Emotion-Detection-with-OpenCV-DeepFace

2. Install dependencies

   pip install -r requirements.txt

3. Download pre-trained model (Optional)

   Download the FER2013 Mini-XCEPTION model from here and place it in the project root directory.

4. Run the application

   python main.py

To train your own models or experiment with the FER2013 dataset:

1. Get Kaggle API credentials

   • Go to Kaggle Account Settings
   • Click "Create New API Token"
   • Save the downloaded kaggle.json to ~/.kaggle/ (Linux/Mac) or C:\Users\\.kaggle\ (Windows)

2. Download the dataset

   python scripts/download_dataset.py

For detailed Kaggle setup instructions, see KAGGLE_SETUP.md.

1. Face Detection: MediaPipe Face Detection (faster and more accurate than Haar Cascades)
2. Emotion Analysis:
   • Primary: Custom FER2013 Mini-XCEPTION CNN
   • Fallback: DeepFace with multiple backend support
3. Performance Optimization:
   • Frame throttling (analyze every N frames)
   • Multithreaded analysis pipeline
   • Efficient NumPy operations

The system recognizes 7 fundamental emotions based on Paul Ekman's research:

• FPS: 60+ on modern CPUs (with GPU: 120+)
• Latency: < 50ms per frame
• Accuracy: ~65-70% on FER2013 test set
• Memory: ~500MB RAM usage

Edit src/config.py to customize:

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

This project is licensed under the MIT License - see the LICENSE file for details.

• DeepFace: serengil/deepface
• FER2013 Dataset: Kaggle FER2013
• MediaPipe: Google MediaPipe
• Mini-XCEPTION: oarriaga/face_classification

Shayan Taherkhani

• Website: shayantaherkhani.ir
• GitHub: @Shayanthn
• Email: admin@shayantaherkhani.ir

• First run may be slow due to model loading
• Requires good lighting for optimal accuracy
• Multiple faces in frame: only the first detected face is analyzed

• Multi-face support
• Age and gender detection
• Emotion history timeline graph
• Export analysis data to CSV/JSON
• Web dashboard for remote monitoring
• Mobile app (iOS/Android)
• Cloud deployment (AWS/Azure)

If you have any questions or need help, please:

1. Check the Issues page
2. Create a new issue with detailed information
3. Contact via email: admin@shayantaherkhani.ir

Made with by Shayan Taherkhani

Star this repository if you find it helpful!