Emotion-Detection

Emotion Detection with CNN

This project uses a Convolutional Neural Network (CNN) built with PyTorch to detect emotions from facial images. The model is trained on the Fer 2013 dataset (32,000+) labelled images. It includes data preprocessing, model training with early stopping, evaluation, and visualization of results.

Features:

• Custom CNN architecture for emotion classification
• Data augmentation and normalization
• Early stopping to prevent overfitting
• Training and test loss visualization
• Webcam-based real-time emotion detection (with OpenCV)
• Easy-to-follow Jupyter notebook

Getting Started

Prerequisites>Python 3.9 & Cuda

See requirements.txt for all dependencies

Installation

1. Clone the repository: git clone https://github.com/temii70/Emotion-Detection.git

2. Install dependencies: pip install -r requirements.txt

3. Download and extract the dataset from https://www.kaggle.com/datasets/msambare/fer2013 into the train and test folders

Usage

• Run model.ipynb in Jupyter Notebook or your IDE to train and evaluate the model
• For real-time emotion detection, run the emotion_script.ipynb provided

Project Structure

• model.ipynb — Main notebook for training and evaluation
• model.py — CNN model definition
• requirements.txt — List of required Python packages
• archive — Folder for training and test images (not included in repo)

This project is licensed under the Apache License.

Demo:

Webcam Live Detection:

Works on multiple faces at the same time:

Next Steps: This is a ongoing project, and as such I will be updating/improving the model and adding other applications as time progresses.