Skip to main content

Lesson 4: Integrating With The Browser (Advanced)

You can use the below button to open this project directly in GitPod. It will set up a Rune playground for you.

Open in Gitpod

In this lesson, we will build a simple React web app that uses a rune for object recognition linked to the webcam.

This tutorial will show you how to import our cdn library to deploy your runes on the web.

Deploy to web#

In the project directory, run:

npm install#

Install dependencies

npx parcel index.html#

Opens a web server on http://localhost:1234.

In gitPod make sure you open the app in a new tab, webcam may not work in an iFrame.


Create a react boilerplate app#

npx create-react-app my-app

Import the library#

Add an index.html file and import the rune library

<div id="app"></div>
<script src=""></script>
<script type="module" src="src/index.tsx"></script>

Write the functions to deploy and run your rune#

// url for the rune
const RUNE_URL = "";
export const runRune = async (image: HTMLImageElement) => {
let input = new Uint8Array();
let output = { elements:""};
let runTime;
//this ImageCapability class links the input data
class ImageCapability {
parameters = {};
generate(dest, id) {
dest.set(input, 0);
setParameter(key, value) {
this.parameters[key] = value;
//this SerialOutput class links the output data
class SerialOutput {
consume(data) {
const utf8 = new TextDecoder();
output = JSON.parse(utf8.decode(data));
const imageCap = new ImageCapability();
//Create all the imports for the Runtime
const imports = {
createCapability: () => imageCap,
createOutput: () => new SerialOutput(),
createModel: (mime, model_data) =>
log: (log) => { console.log(log) },
//download rune and deploy as a Runtime
if (runTime === undefined) {
const response = await fetch(RUNE_URL);
const bytes = new Uint8Array(await response.arrayBuffer());
runTime = await rune.Runtime.load(bytes.buffer, imports);
//get image input and resize
input = rune.TensorFlowModel.resizeImage(image, 224);
// call the model;
//get output and convert to image
console.log("Result:", output);
return output.elements;

Add the Webcam component#

Add the Webcam component for React to your dependencies

npm install react-webcam

Import it in App.tsx

import Webcam from "react-webcam";

Link everything together#

run = async () => {
const imageSrc =;
const img = new Image();
img.src = imageSrc;
const out = await runRune(img);
this.setState({ output: out })