d3.facedetection

d3.js Face Detection Plugin

A d3.js plugin to detect faces on images (both HTML and SVG), videos and canvases to get their coordinates.

Importante note: This plugin uses an algorithm by Liu Liu.

This is a fork of the very popular jquery.facedetection by jaysalvat

Get started

Download the plugin with the method of your choice.

• Download the last release manually
• Or install it with Bower.

bower install d3.facedetection

• Or install it with NPM.

npm install d3.facedetection

Include d3 and the plugin.

<script src="//cdnjs.cloudflare.com/ajax/libs/d3/3.5.3/d3.min.js"></script> 
<script src="path/to/dist/d3.facedetection.min.js"></script> 

Set a picture with some faces in your HTML (or SVG) page.

<svg height="350px" width="620px" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink= "http://www.w3.org/1999/xlink">
  <image id="picture" xlink:href="picture.jpg" x="0" y="0" width="620px" height="350px"></image>
</svg>

Apply the plugin to this image and get the face coordinates.

<script>
    d3.select('#picture').faceDetection({
        complete: function (faces) {
            console.log(faces);
        }
    });
</script> 

Results

Returns an array of found faces object:

• x — Y coord of the face in the picture
• y — Y coord of the face in the picture
• width — Width of the face
• height — Height of the face
• positionX — X position relative to the document
• positionY — Y position relative to the document
• offsetX — X position relative to the offset parent
• offsetY — Y position relative to the offset parent
• scaleX — Ratio between original image width and displayed width
• scaleY — Ratio between original image height and displayed height
• confidence — Level of confidence

Settings

• interval — Interval (default 4)
• minNeighbors — Minimum neighbors threshold which sets the cutoff level for discarding rectangle groups as face (default 1)
• confidence — Minimum confidence (default null)
• async — Async mode if Worker available (default false). The async mode uses Workers and needs the script to be on the same domain.
• grayscale — Convert to grayscale before processing (default true)
• complete — Callback function trigged after the detection is completed

complete: function (faces) {
    // ...
}

• error — Callback function trigged on errors

error: function (code, message) {
    // ...
}