MoveNet Multipose

MoveNet multiple body pose detection from Google MediaPipe.

Installing MoveNet Multipose

Add the following items to your Unity project's Packages/manifest.json:

{
  "scopedRegistries": [
    {
      "name": "NatML",
      "url": "https://registry.npmjs.com",
      "scopes": ["ai.natml"]
    }
  ],
  "dependencies": {
    "ai.natml.vision.movenet-multipose": "1.0.1"
  }
}

Predicting Poses in an Image

First, create the MoveNet Multipose predictor:

// Create the MoveNet Multipose predictor
var predictor = await MoveNetMultiposePredictor.Create();

Then create an input feature:

// Create image feature
Texture2D image = ...;
var input = new MLImageFeature(image);
// Set the normalization
(input.mean, input.std) = modelData.normalization;

Finally, detect the body pose in an image:

// Detect the body pose
MoveNetMultiposePredictor.Pose[] pose = predictor.Predict(input);

Requirements

• Unity 2021.2+

Quick Tips

• Join the NatML community on Discord.
• Discover more ML models on NatML Hub.
• See the NatML documentation.
• Discuss NatML on Unity Forums.
• Contact us at [email protected].

Thank you very much!