Carpet-Defect-Classification

Overview

The Carpet-Defect-Classification project aims to distinguish defective carpets from perfect ones by utilizing the VGG16 model, a widely used convolutional neural network (CNN) architecture. The project implements a modular pipeline for training and deployment, leveraging deep learning techniques for image classification. Additionally, it integrates Continuous Integration (CI) and Continuous Deployment (CD) processes through AWS services and GitHub Actions."

Dataset

The dataset used for this project is sourced from the MVTec Anomaly Detection Dataset, which contains images for defect detection across various categories, including textiles like carpets.

Project Workflow

Steps to run the project:

1. Clone the Repository

   git clone https://github.com/abbykabraham/Carpet-Defect-Classification.git
   cd Carpet-Defect-Classification

2. Create a Conda Environment Create a new Conda environment and activate it:

   conda create -n carpetcls python=3.11 -y
   conda activate carpetcls

3. Install Dependencies Install all the required dependencies using pip:

   pip install -r requirements.txt

4. Run the Application To start the application and model training:

   python app.py

5. Access the Application After running the above command, access the app through the local host and port assigned by Flask.

DVC (Data Version Control) Commands

This project uses DVC for tracking datasets and managing the model training pipeline.

1. Initialize DVC:

   dvc init

2. Run the Pipeline: Reproduce the pipeline stages to train the model and manage data dependencies:

   dvc repro

AWS CI/CD Pipeline with GitHub Actions

Steps for Deployment

1. Login to AWS Console

   • Login to your AWS console and create an IAM user with necessary permissions.

2. AWS Resources Used:

   • EC2 (Elastic Compute Cloud): Used for launching virtual machines.
   • ECR (Elastic Container Registry): For storing Docker images of the application.

3. Deployment Process:

   • Build the Docker Image: Dockerize the application and push the image to Amazon ECR.
   • Launch EC2: Start an EC2 instance to host the application.
   • Pull the Image: Pull the Docker image from ECR into EC2.
   • Run the Container: Deploy and run the Docker container on the EC2 instance.

Policies and Permissions

Ensure the following policies are attached to your IAM user:

• AmazonEC2ContainerRegistryFullAccess: Grants full access to ECR.
• AmazonEC2FullAccess: Grants full access to EC2.

Steps for Docker Setup on EC2

1. Install Docker: Update and install Docker on the EC2 instance:

   sudo apt-get update -y
   sudo apt-get upgrade
   curl -fsSL https://get.docker.com -o get-docker.sh
   sudo sh get-docker.sh
   sudo usermod -aG docker ubuntu
   newgrp docker

2. Setup Self-hosted GitHub Runner:

   • Configure the EC2 instance as a self-hosted runner for GitHub Actions. This will enable CI/CD integration directly on your EC2 instance.
   • Go to GitHub repository settings, and follow the steps under Actions > Runners > New self-hosted runner.

3. Setup GitHub Secrets:

   • Add your AWS credentials and other necessary information in GitHub secrets for secure access during deployment.

Additional Details

• Config Files:
  • config.yaml: Contains configurations related to data paths, model preparation, callback settings, and training parameters.
  • params.yaml: Defines hyperparameters for model training.