Welcome to Car Detection Mlops’s documentation!

Deploy a full Mlops Object detection Web application that detects cars and motobikes!

Features:

• Download and preprocess the Stanford Cars Dataset.

• Data versioning with DVC.

• Train model with YOLO and export in ONNX format.

• AWS API Gateway is created using an AWS Lambda function inside an AWS ECR that make predictions using the trained model.

• AWS S3 Bucket used for storing Dataset versions and the model that will be used by the Lambda function.

• Documentation created using Sphinx and deployed using Github Pages.

• Web Application deployed using Heroku.

• Automated pipeline to deploy all the infrastructure to production using Github Actions.

• Local logging and also CloudWatch logging (for the lambda function).

Tutorial:

• Startup
• Steps for data versioning
  • Create a new data enviroment
  • Create a new dataset version
  • Use specific data version
• Steps for training
• Steps for deploying

Modules:

• Train module
  • main()
  • train_with_YOLO()
• Predict module
  • make_predictions()
• Dataset module
  • Dataset
    • Dataset.dowload_dataset()
    • Dataset.merge_train_test()
    • Dataset.move_files()
    • Dataset.preprocess()
• S3 Bucket module
  • S3_Bucket
    • S3_Bucket.add_file_model_S3_bucket()
    • S3_Bucket.create_S3_bucket()
    • S3_Bucket.delete_S3_bucket()
    • S3_Bucket.handle_arguments()
    • S3_Bucket.list_S3_buckets()
• Lambda Function module
  • intersectionOverUnion()
  • make_prediction()
• Deploy API module
  • Deployment
    • Deployment.create_API()
    • Deployment.create_lambda_function()
    • Deployment.full_deploy()
• Logs module
  • get_logs()

Scripts:

• Bash scripts
  • configure_dvc.sh
  • data.sh
  • new_dataset_version.sh
• YML script
  • workflow.yml
  • workflow_docs.yml

Indices and tables

• Index

• Module Index

• Search Page