Project Title: Monitoring Employee's Emotional Well-being with GoogleNet Using the FER2013 Dataset

Overview

This project utilizes a pre-trained neural network, GoogleNet, to monitor employees' emotional well-being by analyzing facial expressions. The system leverages the FER2013 dataset, containing 30,000 labelled grayscale facial images, to classify emotions such as anger, disgust, fear, happiness, sadness, surprise, and neutrality.

Objectives

• Implement emotion detection using GoogleNet and FER2013 dataset.
• Develop a method to monitor employees' emotional states over time.
• Analyze emotional data to identify trends and patterns in employee well-being.

Key Components

• Dataset: FER2013, preprocessed and augmented.
• Neural Network: GoogleNet with transfer learning.
• Programming Language: MATLAB

Project Structure

1. Introduction: Objectives and overview.
2. Simulations: Details on the FER2013 dataset, encoding, and GoogleNet architecture.
3. Model Evaluation: Accuracy, accuracy curve, and confusion matrix.
4. Critical Analysis: Evaluation of results and suggestions for improvement.
5. Conclusions: Key takeaways and implications.
6. References: Cited works.
7. Certificates: MATLAB course certificates.

How to Run

1. Preprocess the FER2013 dataset.
2. Use MATLAB to load and modify GoogleNet.
3. Train the network with the provided code.
4. Evaluate the model performance using the validation dataset.

Dependencies

• MATLAB
• Deep Learning Toolbox in MATLAB