Creating a Smart and Serverless Cat Detection System

Using Amazon Web Services!

Linda Nichols

@lynnaloo

Norfolk.js

norfolkjs.org

RevolutionVA

revolutionconf.com

hrdevfest.org

Let's Talk about Cat Detection Systems!

@lynnaloo

What is it?

@lynnaloo

A motion detector that sends an SMS message when a cat is nearby.

Components

@lynnaloo

  • Raspberry Pi 3

  • Raspberry Pi Camera

  • PIR Sensor

  • Power Source

  • Lead Wires

  • 1 Kitty (optional)

Hardware

@lynnaloo

Software

@lynnaloo

  • Node.js

  • Johnny-Five

  • AWS IoT SDK

  • Raspicam

  • Node.js

  • Twilio

On the Raspberry Pi:

In the Cloud:

What's Johnny-Five?

@lynnaloo

Johnny-Five

@lynnaloo

Open-source JavaScript Robotics Platform

 

www.johnny-five.io

What is Amazon IoT?

@lynnaloo

It's a managed cloud platform that lets connected devices easily and securely interact with cloud applications and other devices.

- Amazon Web Services

What is Amazon IoT?

@lynnaloo

What is Amazon IoT?

@lynnaloo

It's a pub/sub system for devices.

- Linda

What is Amazon IoT?

@lynnaloo

  • Supports billions of devices

  • Supports trillions of messages

  • Inexpensive

  • Reliable

  • Secure

  • Super-fast

@lynnaloo

Trillions of cat detections

How Does It Work?

@lynnaloo

  • Major components are the device gateway and rules engine.

  • A message broker provides a pub/sub via MQTT protocol.

  • The rules engine filters and routes messages to other services.

How Does It Work?

@lynnaloo

But I thought this was "serverless"... 

@lynnaloo

Disclaimer:

"Serverless" is just a catchy, made-up, marketing term.

@lynnaloo

Serverless Architecture

It's about the events!

@lynnaloo

Serverless Architecture

Events from IoT (and other services) can trigger Serverless Functions 

@lynnaloo

Serverless Functions (FaaS).

@lynnaloo

Stateless functions running in containers managed by a service provider (not you!).

AWS Lambda

  • Very Inexpensive

  • Easy Deployments

  • Scalable

  • Supports Many Runtimes

@lynnaloo

Lambda Handler

@lynnaloo
'use strict';

const catNames = require('cat-names');
const AmazonDb = require('amazon-db-module');

function getCatName(event, context, callback) {
    const amazonDb = new AmazonDb();
    const catName = catNames.random();
    amazonDb.update(catName);
    callback(null, { payload: `Your cat name is ${catName}.` });
}

module.exports.getCatName;

But what makes the cat detector... smart?

@lynnaloo

Amazon Rekognition

@lynnaloo

Features

  • Object and Scene Detection

  • Image Moderation

  • Facial Recognition

  • Face Comparison

  • Celebrity Recognition

Amazon Rekognition

@lynnaloo

Amazon Rekognition service uses deep learning algorithms to analyze images for scenes, objects, faces, and other content.

Visual Recognition knows if a cat is in an image.

Amazon Rekognition

@lynnaloo
[ 
    { Name: 'Animal', Confidence: 98.29473114013672 },
    { Name: 'Cat', Confidence: 97.9908218383789 },
    { Name: 'Adorable', Confidence: 97.9908218383789 },
    { Name: 'Feline', Confidence: 96.9802474975586 },
    { Name: 'Domestic Cat', Confidence: 64.9802474975586 },
    { Name: 'Mammal', Confidence: 64.9908218383789 },
    { Name: 'Manx', Confidence: 64.9908218383789 },
    { Name: 'Pet', Confidence: 64.9908218383789 } 
]

The Code

@lynnaloo

Hardware/IoT Platform Code: github.com/lynnaloo/kitty-detector

AWS Lambda Functions:  github.com/lynnaloo/go-away-kitty

Demo Time!

@lynnaloo

Thank You!

@lynnaloo

Made with Slides.com