I trained a pose model on teachablemachine.with google.com. I tried to use that model as a web-application, so I applied it to the ml5.js on JavaScript like below.

const imageModelURL = 'teachable machine model URL';

async function start() {
  const stream = await navigator.mediaDevices.getUserMedia({
    audio: false,
    video: true,
  });
  
  const video = document.getElementById('webcam');
  
  video.srcObject = stream;
  
  classifier = ml5.imageClassifier(imageModelURL + 'model.json', video, () => {
    console.log('モデルの読み込みが完了しました');
  });
  
  function loop() {
    classifier.classify(async (err, results) => {
      if (results[0]) {
        document.getElementById("result").textContent = results[0].label;// 
      }
      setTimeout(loop, 1000);
    });
  }
  loop();
}

I can use this on other models made by theachable machine image project, but it doesn't work in the case of pose model. I red the article that normally we can't use the pose model on ml5.js, but is there any way to use?