# Pose

Pose is a pre-trained model on EdgeAI Studio (opens new window) for 2D human pose estimation.

Pose estimation is a computer vision technique that detects human figures in images and video, for example, where someone's eye, shoulder, knee, etc... shows up in an image. It is not about recognising who is in the image, but simply about predicting where key body joints are.

The model is able to predict up to 17 different body joint (or key point) coordinates which collectively form a human pose including:

LShoulder LElbow LWrist LHip LKnee LAnkle LEye LEar Nose
RShoulder RElbow RWrist RHip RKnee RAnkle REye REar

pose

Key terms:

  • Pose: returns an object containing a list of keypoints for each detected person.
  • Keypoint: returns an object containing the 2D x, y coorindates and confidence level score of each key body part/joint of a person's pose estimation i.e. nose, right shoulder, left hip, etc.
"output": [
    {
        "Nose": {
          "x": 680.9286671539961,
          "y": 114.56444170321639,
          "confidence": 0.86328125
        },
        "LEar": {
          "x": 682.7241715399609,
          "y": 209.61074561403507,
          "confidence": 0.75634765625
        },
        ...
    },
    {
        ...
    }, 
    ...
]

To use the Pose block for pose detection, you should set up a pipeline like below:

Pose pipeline example

You may find a demo pipeline here (opens new window).