Deep Facial Emotion Detection (CNN)

Table of Contents

• Overview
• Features
• Installation
• Usage
• Acknowledgements

Overview

• This project is an emotion classification application built using the FER2013 dataset.
• It employs a convolutional nerual network to predict emotions from facial images.
• The application is built using TensorFlow and Streamlit, making it a full-stack deep learning project.

Features

• Emotion Prediction: Classify the emotion of a given facial image (e.g., happy, sad, angry).
• Webcam Capture: Capture images directly using your webcam.
• File Upload: Upload an image file for emotion classification.

Performance

• The image below showcases the performance numbers achieved for the different models within this project (current world record = 75%)

=

Installation

To run this application locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/mohammed-majid/CNN_emotion_classification.git
   

2. Install the required packages:

   pip install -r requirements.txt
   

3. Download the pre-trained model and place it in the project directory:

   • custom_model_v2.h5

4. Run the Streamlit application:

   streamlit run app.py
   

   or

   python3 -m streamlit run app.py
   

Usage

1. Open the Streamlit application in your web browser.

2. Choose between using the webcam or uploading an image file:

   • Webcam: Click the "Capture Image" button to take a picture.
   • File Upload: Click the "Upload Image" button to upload a file from your computer.

3. Click the "Predict Emotion" button to get the emotion prediction.

Acknowledgements

This project was developed using the following libraries and tools:

• TensorFlow
• Keras
• OpenCV
• Streamlit
• NumPy
• Pandas
• Scikit-learn
• Matplotlib

Side Note

• In case you want to check the dataset out, Press here.