Hair Matte Net

Hair matte segmentation. This predictor implements HairMatteNet. This package requires NatML.

Segmenting Hair in an Image

First, create the predictor:

// Fetch model data from NatML Hub
var modelData = await MLModelData.FromHub("@natsuite/hair-matte-net");
// Deserialize the model
var model = modelData.Deserialize();
// Create the hair segmentation predictor
var predictor = new HairMatteNetPredictor(model);

Then segment hair in an image:

Texture2D image = ...; // Can also be an `MLImageFeature`
// Segment the hair in the image
HairMatteNetPredictor.Matte matte = predictor.Predict(image);

Finally, render the segmentation map to a RenderTexture:

// Create a `RenderTexture`
var hairMask = new RenderTexture(image.width, image.height, 0);
// Render the segmentation map into the render texture
matte.Render(hairMask);

References

• https://arxiv.org/pdf/1712.07168.pdf
• https://github.com/wonbeomjang/mobile-hair-segmentation-pytorch

Requirements

• Unity 2020.3+
• NatML 1.0.16+

Quick Tips

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

Thank you very much!