@api.video/media-stream-person-segmentation
v0.0.1
Published
api.video media stream person segmentation - change the background and emphasize the person in your media streams
Downloads
9
Readme
api.video is the video infrastructure for product builders. Lightning fast video APIs for integrating, scaling, and managing on-demand & low latency live streaming features in your app.
Table of contents
Project description
This library allows you to easily blur the background of a webcam video stream in real-time using TensorFlow.
With this library, you can create a more professional and visually appealing video conferencing experience by keeping the focus on the person in front of the camera and reducing visual distractions from the surrounding environment.
Please note that this project is currently experimental and may not be suitable for production use. This works pretty well on Chrome and Firefox, it may not work as well on Safari.
Getting started
Installation
Simple include in a javascript project
Include the library in your HTML file like so:
<head>
...
<script src="https://unpkg.com/@api.video/media-stream-person-segmentation" defer></script>
</head>...
<script type="text/javascript">
navigator.mediaDevices.getUserMedia(constraints).then((webcamStream) => {
const segmentation = new ApiVideoMediaStreamPersonSegmentation(webcamStream)
segmentation.applyBlurEffect()
segmentation.onReady((ouputStream) => {
// use the ouputStream to display the video
// ...
});
});
</script>Documentation
Instanciation
Options
The library is instanciated with a MediaStream object as parameter.
const segmentation = new ApiVideoMediaStreamPersonSegmentation(webcamStream)Methods
applyBlurEffect(options: BlurEffectOptions)
Apply a blur effect on the background of the video stream.
Options
| Name | Default | Type | Description | | :------ | :------| :------ | :------ | | backgroundBlurAmount | 15 | number | The amount of blur to apply to the background. | | edgeBlurAmount | 8 | number | The amount of blur to apply to the edges of the person. |
segmentation.applyBlurEffect({
backgroundBlurAmount: 15,
edgeBlurAmount: 8
})onReady(callback: (stream: MediaStream) => void)
Call the callback when the stream is ready.
segmentation.onReady((stream) => {
// use the stream to display the video
// ...
});Full example
<html>
<head>
<script src="https://unpkg.com/@api.video/media-stream-person-segmentation"></script>
<style>
#container {
display: flex;
flex-direction: column;
align-items: center;
}
#video {
width: 640;
height: 480;
border: 1px solid gray;
}
#container div {
margin: 10px 0;
}
label {
display: inline-block;
width: 160px;
text-align: right;
}
input {
display: inline-block;
width: 300px;
}
</style>
</head>
<body>
<div id="container">
<div>
<video muted id="video"></video>
</div>
<div>
<div>
<label for="blur">Blur background</label>
<input type="range" id="blur" min="0" max="100" value="15" />
</div>
<div>
<label for="blurEdges">Blur edges</label>
<input type="range" id="blurEdges" min="0" max="10" value="8" />
</div>
</div>
</div>
<script>
const video = document.querySelector('#video');
var constraints = window.constraints = {
audio: true,
video: true
};
navigator.mediaDevices.getUserMedia(constraints).then((stream) => {
const segmentation = new ApiVideoMediaStreamPersonSegmentation(stream)
segmentation.applyBlurEffect({
backgroundBlurAmount: 15,
edgeBlurAmount: 8
})
document.getElementById("blur").addEventListener("input", (e) => segmentation.applyBlurEffect({
backgroundBlurAmount: e.target.value
}));
document.getElementById("blurEdges").addEventListener("input", (e) => segmentation.applyBlurEffect({
edgeBlurAmount: e.target.value
}));
segmentation.onReady((stream) => {
video.srcObject = stream;
video.play();
});
});
</script>
</body>
</html>