Facial Emotion Recognition 😃

This project implements a real-time facial emotion detection system using deep learning techniques. The model is trained to recognize emotions from facial expressions in images or video streams.

Table of Contents 📚

• Cloning the Repository
• Dataset
• Setting Up the Virtual Environment
• Requirements
• Folder Structure
• Preparing the Dataset
• Training the Model
• Real-time Emotion Detection
• Haarcascade Frontal Face Detection

Cloning the Repository 🧑‍💻

To get started, clone the repository using the following command:

git clone https://github.com/VinamraSaurav/FacialEmotionRecognition
cd FacialEmotionRecognition

Dataset 📊

The project includes the fer2013.csv dataset, which contains facial expressions labeled with their corresponding emotions. This dataset is crucial for training the emotion detection model.

Download the fer2013.csv dataset from sources like Kaggle and save it to the root directory of this project.

Setting Up the Virtual Environment 🛠️

To set up a Python virtual environment, follow these steps:

1. Create a Virtual Environment:

   python -m venv venv

2. Activate the Virtual Environment:

   • On Windows:

     venv\Scripts\activate

   • On macOS/Linux:

     source venv/bin/activate

3. Install Requirements: Install the required packages using the requirements.txt file:

   pip install -r requirements.txt

Requirements 📦

The following packages are required to run this project:

• TensorFlow
• Keras
• OpenCV
• NumPy
• Pandas
• Matplotlib

These dependencies are specified in the requirements.txt file, and you can install them using the above command.

Folder Structure 🗂️

The project has the following folder structure:

FacialEmotionRecognition/
│
├── src/
│   ├── dataset_prepare.py          # Script to prepare the dataset from CSV
│   ├── emotions.py                 # Real-time emotion detection script
│   ├── train.py                    # Script for training the model
│   └── haarcascade_frontalface_default.xml  # Haar Cascade file for face detection
│
├── data/                           # Directory containing folders for each emotion
│   ├── angry/                      # Folder for images of angry faces
│   ├── sad/                        # Folder for images of sad faces
│   ├── happy/                      # Folder for images of happy faces
│   └── ...                         # Additional emotion folders
│
├── venv/                           # Python virtual environment directory
│
├── model.h5                        # Saved trained model file
├── fer2013.csv                     # Dataset in CSV format
├── README.md                       # Project documentation file
└── requirements.txt                # List of dependencies

Preparing the Dataset 🗃️

Once you have the 'fer2013.csv' in your root directory, use the dataset_prepare.py script to process the CSV file, which will organize the images into folders based on emotions.

Run the following command to prepare the dataset:

python src/dataset_prepare.py

This script reads fer2013.csv in root directory and creates emotion-specific folders with the images organized for model training. Make sure fer2013.csv is stored in the root directory of the project before running this command.

🚨 ** NOTE ** - If you plan to use a different CSV file, please ensure you update the `data_dir` variable in `dataset_prepare.py` to reflect the correct file name and location ! 📂💡

Training the Model 🚀

To train the model using the prepared dataset, you can use the train.py script. Ensure that the dataset is prepared before running this script. Run the following command:

python src/train.py

This will start the training process and save the trained model as model.h5.

Real-time Emotion Detection 🎥

The emotions.py script is used for real-time emotion detection. It captures video from the webcam and predicts emotions based on the faces detected in the video stream. Run the script using:

python src/emotions.py

Make sure the trained model (model.h5) is present in the root directory.

Haarcascade Frontal Face Detection 👀

The project uses haarcascade_frontalface_default.xml, a pre-trained Haar Cascade model, for detecting faces in images and video streams. This XML file is included in the project directory and is used within the emotions.py script to identify faces before predicting their emotions.