Deep Learning Masterclass 1

Introduction

This repository contains the code for thefinal deliverable of the Deep Learning Masterclass 1 Hildesheim Universität. It was built using the fastai tutorial for AWS deployment and the AWS tutorial on SAM

Repository structure

The repository contains the following folders:

• front-end: contains a working example of a web-client built with Bootstrap.
• obj-detector: code of the object detector project. NOTE: up until 01-10-2019, the code only work in a local environment (Docker Desktop for Windows)
• sam-app-backup: some failed experiments on the implementation of the project
• sam-app: working example of a web service using AWS SAM. receives an encoded base64 image and returns the that same image flipped 90 degrees

Build instructions

sam-app

To build the project sam-app, clone the repo and move to that folder

cd sam-app

Then run the following commands

sam build

sam package --output-template packaged.yaml --s3-bucket {NAME OF YOUR AWS BUCKET}

sam deploy --template-file packaged.yaml --region {REGION TO DEPLOY} --capabilities CAPABILITY_IAM --stack-name aws-sam-getting-started

obj-detector

NOTE: the deployment only worked for local testing

cd obj-detector

sam build --use-container

sam package --output-template packaged.yaml --s3-bucket {NAME OF YOUR AWS BUCKET}

Test locally

sam local start-api

Following comand throws an error sam deploy --template-file packaged.yaml --region {REGION TO DEPLOY} --capabilities CAPABILITY_IAM --stack-name pytorch-sam-app