UNPKG

scrawl-canvas

Version:
57 lines (45 loc) 1.92 kB
<!DOCTYPE html> <html> <head> <meta charset="utf-8"> <meta http-equiv="x-ua-compatible" content="ie=edge"> <meta name="viewport" content="width=device-width, initial-scale=1"> <title>Demo Tensorflow 001</title> <link href="css/normalize.css" rel="stylesheet" /> <link href="css/tests.css" rel="stylesheet" /> <!-- Examples all show the files being loaded in the head, not at the end of the body --> <!-- Load TensorFlow.js - example uses v1.2, not v2 --> <script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@1.2"></script> <!-- Load BodyPix --> <script src="https://cdn.jsdelivr.net/npm/@tensorflow-models/body-pix@2.0"></script> </head> <body> <h1><a href="index.html">Scrawl-canvas v8</a> - Tensorflow test 001</h1> <h2>Tensorflow tfjs-models / body-pix experiment - follow my eyes</h2> <h3>This demo should work in Chrome and Firefox browsers; fails in Safari</h3> <canvas id="mycanvas" width="600" height="400" data-scrawl-canvas ></canvas> <p id="reportmessage"></p> <div class="testinfo"> <h4>Test purpose</h4> <ul> <li>Check that the media stream loads (if browser/device supports it)</li> <li>Check that we can load and use both Tensorflow (from a CDN) and its body-pix model</li> <li>Apply the model to a live camera feed</li> <li>Create a RawAsset object to handle the model's output. In particular</li> <ul> <li>Parse the output's metadata to identify the left and right eyes</li> <li>Calculate the position of each eye, averaged over several iterations</li> <li>Update the RawAsset's canvas with two transluscent circles covering over the eyes</li> </ul> </ul> <p><b>Touch test:</b> not required</p> <p><a href="../docs/demo/tensorflow-001.html">Annotated code</a></p> </div> <script src="tensorflow-001.js" type="module"></script> </body> </html>