관리 메뉴

웹개발자의 기지개

[Deep Learning] Teachable Machine - 실시간 탐지하기 본문

인공지능/딥러닝

[Deep Learning] Teachable Machine - 실시간 탐지하기

http://portfolio.wonpaper.net 2023. 11. 6. 21:22

 

캠코더나 이미지로 실시간 전이학습을 이용한 테스트가 가능하다.

 

https://teachablemachine.withgoogle.com

 

Teachable Machine

Train a computer to recognize your own images, sounds, & poses. A fast, easy way to create machine learning models for your sites, apps, and more – no expertise or coding required.

teachablemachine.withgoogle.com

 

 

 

 

간단히 html 파일을 만들어 돌리면

 

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
<div>Teachable Machine Image Model</div>
<button type="button" onclick="init()">Start</button>
<div id="webcam-container"></div>
<div id="label-container"></div>
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@latest/dist/tf.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/@teachablemachine/image@latest/dist/teachablemachine-image.min.js"></script>
<script type="text/javascript">
    // More API functions here:
    // https://github.com/googlecreativelab/teachablemachine-community/tree/master/libraries/image
 
    // the link to your model provided by Teachable Machine export panel
    const URL = "https://teachablemachine.withgoogle.com/models/hm5drTELE/";
 
    let model, webcam, labelContainer, maxPredictions;
 
    // Load the image model and setup the webcam
    async function init() {
        const modelURL = URL + "model.json";
        const metadataURL = URL + "metadata.json";
 
        // load the model and metadata
        // Refer to tmImage.loadFromFiles() in the API to support files from a file picker
        // or files from your local hard drive
        // Note: the pose library adds "tmImage" object to your window (window.tmImage)
        model = await tmImage.load(modelURL, metadataURL);
        maxPredictions = model.getTotalClasses();
 
        // Convenience function to setup a webcam
        const flip = true// whether to flip the webcam
        webcam = new tmImage.Webcam(200200, flip); // width, height, flip
        await webcam.setup(); // request access to the webcam
        await webcam.play();
        window.requestAnimationFrame(loop);
 
        // append elements to the DOM
        document.getElementById("webcam-container").appendChild(webcam.canvas);
        labelContainer = document.getElementById("label-container");
        for (let i = 0; i < maxPredictions; i++) { // and class labels
            labelContainer.appendChild(document.createElement("div"));
        }
    }
 
    async function loop() {
        webcam.update(); // update the webcam frame
        await predict();
        window.requestAnimationFrame(loop);
    }
 
    // run the webcam image through the image model
    async function predict() {
        // predict can take in an image, video or canvas html element
        const prediction = await model.predict(webcam.canvas);
        for (let i = 0; i < maxPredictions; i++) {
            const classPrediction =
                prediction[i].className + ": " + prediction[i].probability.toFixed(2);
            labelContainer.childNodes[i].innerHTML = classPrediction;
        }
    }
</script>
 
cs

 

 

 

Comments