Emotion Recognition Model Using CNN

This project involves developing a Convolutional Neural Network (CNN) for recognizing human emotions from facial expressions using the FER2013 dataset.

Overview

The model classifies images into one of seven emotions: Angry, Disgust, Fear, Happy, Sad, Surprise, and Neutral. It leverages advanced data preprocessing techniques and regularization strategies to optimize performance and generalization.

Features

• Emotion Classification: Classifies images into seven emotion categories.
• Data Preprocessing: Includes normalization and augmentation for enhanced model robustness.
• Regularization: Uses Batch Normalization and Dropout to enhance model performance.
• Model Training: Built and trained using TensorFlow and Keras.

Dataset

The FER2013 dataset is required for training and evaluation. It is not included in this repository due to its large size.

You can download the dataset from Kaggle’s FER2013 page.

After downloading, place the dataset in the dataset/FER2013 directory with the following structure:

dataset/
└── FER2013/
    ├── train/
    │   ├── angry/
    │   ├── disgust/
    │   ├── fear/
    │   ├── happy/
    │   ├── sad/
    │   ├── surprise/
    │   └── neutral/
    └── test/
        ├── angry/
        ├── disgust/
        ├── fear/
        ├── happy/
        ├── sad/
        ├── surprise/
        └── neutral/

Installation

1. Clone the repository:

   git clone https://github.com/your-username/emotion-recognition-model.git

2. Navigate to the project directory:

   cd emotion-recognition-model

3. Set up a virtual environment and install the required packages:

   python -m venv facenv
   source facenv/bin/activate  # On Windows use `facenv\Scripts\activate`
   pip install -r requirements.txt

Usage

1. Ensure the FER2013 dataset is in the dataset/FER2013 directory.

2. Run emotion detection:

   python recognition.py 

3. Explore the notebook:

   jupyter notebook recognition.ipynb

4. Model file: emotion_detection.keras should be in the project directory for the scripts to function.

Results

The model effectively classifies emotions from facial expressions and demonstrates practical insights into emotion recognition. The achieved performance is suitable for applications requiring emotion understanding.

Contributing

Feel free to submit issues or pull requests. Contributions are welcome!

License

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

Acknowledgments

•	FER2013 dataset by Kaggle.
•	TensorFlow and Keras for their powerful libraries.