Home > AWS, re:Invent, Serverless > AWS re:Invent 2018: Build a Photo-Sharing App with AI-Powered Face and Object Detection – MOB306

AWS re:Invent 2018: Build a Photo-Sharing App with AI-Powered Face and Object Detection – MOB306

November 29th, 2018

Dustin Noyes & Gabriel Hollombe from AWS


Another workshop, so my fingers have been getting a workout today, and no easy passive listening to speaker sessions for me!

This workshop was another one I didn’t have any previous knowledge of and seemed interesting to learn something new like my previous session: AWS re:Invent 2018: Get Started with Deep Learning and Computer Vision Using AWS DeepLens – AIM316

You can play along at home with good instructions https://amplify-workshop.go-aws.com

We had to build a data-driven web app using React that lets users upload to shared, secure photo galleries. <sidebar>I’m struggling enough with photo management at home as it is, multiple people, too many photos and complicated ways to sync between devices, back them up and make them available to view! </sidebar>

imageAWS Amplify was used which is a service to create and configure AWS services, connecting front-end application to cloud services like databsaes, Lambda, storage analytics etc. I also learned about AWS AppSync which is a serverless (yay!) back-end for mobile, web, and enterprise applications. AppSync uses GraphQL which is an API language that we used to back up our data in DynamoDB.

We then went into AWS Amplify which is a JavaScript library and we used part of it to authenticate users, communicate with our API, and manage photo uploads. Cognito was used for Authentication.

Of course, we needed some AI so we then tacked on Amazon Rekognition to do some object tagging so users could find photos more easily.

We used Cloud9 as the IDE which I’ve used for most workstations, very simple and impressive to use.

It was a very developer heavy workshop with JS and React specifically which I haven’t used so that took my out of my comfort zone, exactly why I’m here. I managed to mess up one of the early steps by running the installs from the wrong directory so had to start again as there were a bunch of error codes, wasted some time. Annoyingly something else bombed out (likely my fault) just before we got to the AI part, so I’ll have to continue at home.

This was a very well run workshop, we went through the steps with the presenters on the big screen as well, documentation was very well done so no-one was lost. Foolproof docs are so important.

Comments are closed.