ImageSegmentation

in Lib.io.video.mediapipe

Segments an image using MediaPipe Image Segmentation.

This operator uses MediaPipe Image Segmentation task to partition input image into different regions (e.g., background, person, hair, clothing). It outputs segmentation masks and data that can be used for compositing or analysis.

Inputs: - InputTexture: The image to process. - Enabled: Enables or disables image segmentation. - Model: Selects the segmentation model to use (e.g., SelfieSegmenter, DeepLabV3). - SelectedCategories: Comma-separated list of category indices to include in the mask. - Debug: Enables debug logging and debug texture output. - CategoryAllowlist: Comma-separated list of category names to allow (e.g. "person, cup").

Outputs: - OutputTexture: The input texture (passed through). - MaskTexture: A texture where each pixel value corresponds to a category index. - UpdateCount: Increments each time a new frame is processed. - Confidence: The confidence score of the segmentation. - CategoryMask: A texture where each pixel value corresponds to a category index. - ConfidenceMask: A texture containing confidence scores for each pixel. - DebugTexture: Debug visualization texture.

Input Parameters

Name (Relevancy & Type)	Description
InputTexture (Texture2D)	The image to process.
Enabled (Boolean)	Enables or disables image segmentation.
Model (Int32)	Selects the segmentation model to use (e.g., SelfieSegmenter, DeepLabV3).
SelectedCategories (String)	Comma-separated list of category indices to include in the mask.
Debug (Boolean)	Enables debug logging and debug texture output.
CategoryAllowlist (String)	Comma-separated list of category names to allow (e.g. "person, cup").

Outputs

Name	Type
OutputTexture	T3.Core.DataTypes.Texture2D
MaskTexture	T3.Core.DataTypes.Texture2D
UpdateCount	System.Int32
Confidence	System.Single
CategoryMask	T3.Core.DataTypes.Texture2D
ConfidenceMask	T3.Core.DataTypes.Texture2D
DebugTexture	T3.Core.DataTypes.Texture2D