@tensorflow-models/face-landmarks-detection

Face Landmarks Detection

This package provides models for running real-time face detection and landmark tracking.

Currently, we provide 1 model option:

MediaPipe:

Demo

MediaPipe Facemesh can detect multiple faces, each face contains 478 keypoints.

More background information about the package, as well as its performance characteristics on different datasets, can be found here: Model Card. The facemesh package optionally loads an iris detection model, whose model card can be found here: Model Card.

---

Table of Contents

1. How to Run It
2. Keypoint Diagram
3. Example Code and Demos

---

How to Run It

In general there are two steps:

You first create a detector by choosing one of the models from SupportedModels, including MediaPipeFaceMesh.

For example:

const model = faceLandmarksDetection.SupportedModels.MediaPipeFaceMesh;
const detectorConfig = {
  runtime: 'mediapipe', // or 'tfjs'
  solutionPath: 'https://cdn.jsdelivr.net/npm/@mediapipe/face_mesh',
}
const detector = await faceLandmarksDetection.createDetector(model, detectorConfig);

Then you can use the detector to detect faces.

const faces = await detector.estimateFaces(image);

The returned face list contains detected faces for each faces in the image. If the model cannot detect any faces, the list will be empty.

For each face, it contains a bounding box of the detected face, as well as an array of keypoints. MediaPipeFaceMesh returns 478 keypoints. Each keypoint contains x, y and z, as well as a name.

Example output:

[
  {
    box: {
      xMin: 304.6476503248806,
      xMax: 502.5079975897382,
      yMin: 102.16298762367356,
      yMax: 349.035215984403,
      width: 197.86034726485758,
      height: 246.87222836072945
    },
    keypoints: [
      {x: 406.53152857172876, y: 256.8054528661723, z: 10.2, name: "lips"},
      {x: 406.544237446397, y: 230.06933367750395, z: 8},
      ...
    ],
  }
]

The box represents the bounding box of the face in the image pixel space, with xMin, xMax denoting the x-bounds, yMin, yMax denoting the y-bounds, and width, height are the dimensions of the bounding box.

For the keypoints, x and y represent the actual keypoint position in the image pixel space. z represents the depth with the center of the head being the origin, and the smaller the value the closer the keypoint is to the camera. The magnitude of z uses roughly the same scale as x.

The name provides a label for some keypoint, such as 'lips', 'leftEye', etc. Note that not each keypoint will have a label.

Refer to each model's documentation for specific configurations for the model and their performance.

MediaPipeFaceMesh MediaPipe Documentation

MediaPipeFaceMesh TFJS Documentation

---

Keypoint Diagram

See the diagram below for what those keypoints are and their index in the array.

MediaPipe FaceMesh Keypoints

---

Example Code and Demos

You may reference the demos for code examples. Details for how to run the demos are included in the demos/ folder.