Project Overview

This project uses Convolutional Neural Networks (CNNs) to predict age ranges, gender, and emotions from human facial images. The model is trained on popular datasets and deployed using a Flask web application, allowing users to upload images for real-time predictions.

Features

• Age Prediction: Predicts the age range from a facial image.
• Gender Classification: Classifies the image as male or female.
• Emotion Detection: Recognizes common emotions like happy, sad, and neutral.
• Web Application: Provides a user-friendly interface for uploading images and viewing predictions.

Project Structure

• .github/workflows/main.yaml: GitHub Actions workflow for CI/CD, handling code linting, testing, Docker image building, and deployment.
• templates/: Contains HTML files (index.html for image uploads, result.html for displaying predictions).
• app.py & main.py: Flask application files, loading the models and processing images for predictions.
• deployment-steps.txt: Instructions for deploying the application on AWS using EC2 and ECR.
• Dockerfile: Defines a Docker image with the application environment.
• requirements.txt: Lists the dependencies needed for the project.

Dataset

The model is trained using three datasets:

• UTK Face Dataset: Used for age and gender prediction.
• Facial-Age Dataset: Provides labeled age data.
• CT+ Dataset: Used for emotion detection.

Preprocessing

Images are resized and converted to grayscale before inputting into the CNN models:

• Age Model: 200x200 pixels
• Gender Model: 100x100 pixels
• Emotion Model: 48x48 pixels

Model Architecture

The CNN architecture used for this project includes:

1. Multiple Conv2D layers with increasing filter sizes (32, 64, 128, 256).
2. MaxPooling layers following each Conv2D layer to reduce dimensionality.
3. A fully connected Dense layer.
4. Output layer with softmax activation for classifying emotions, age range, and gender.

Performance Metrics

• Age Detection Accuracy: 82%
• Gender Classification Accuracy: 90%
• Emotion Detection Accuracy: 97%

Usage

Running the Project Locally

1. Install Requirements:

   pip install -r requirements.txt

2. Run the Flask App:

   python main.py

3. Access the Application: Open your browser and navigate to http://localhost:5000

Docker Deployment

1. Build Docker Image:

   docker build -t age-gender-emotion-app .

2. Run Docker Container:

   docker run -p 80:80 age-gender-emotion-app

Deployment Guide (AWS)

Follow these steps for AWS deployment:

Create IAM User: Set up permissions for EC2 and ECR.
Create ECR Repository: Store the Docker image.
Launch EC2 Instance: Install Docker on EC2 and configure it as a GitHub runner.
Push Docker Image to ECR: Build and push the image for deployment.
Run Docker on EC2: Pull and run the image to serve the application.
For detailed steps, refer to deployment-steps.txt.

GitHub Actions CI/CD

The project uses GitHub Actions to automate testing and deployment:

Continuous Integration: Lints code and runs unit tests.
Continuous Delivery: Builds and pushes Docker image to ECR.
Continuous Deployment: Deploys to EC2 using a self-hosted runner.

Dependencies

The project requires the following packages:

Flask
TensorFlow
Keras
Pillow
NumPy
OpenCV