Module segmentation 

Available on crate feature segmentation only.

Segmentation mask generation — VNGeneratePersonSegmentationRequest and VNGenerateForegroundInstanceMaskRequest.

Both produce a grayscale CVPixelBuffer mask; this module copies it into a Rust-owned Vec<u8> so callers don’t need a CoreVideo dependency. Mask values are 8-bit (0 = background, 255 = foreground). For instance masks, pixel values index into a list of detected instances (1..=instance_count).

Structs

InstanceMask

A foreground-instance mask plus the number of distinct instances the model identified. Pixel values are 0 for background and 1..=instance_count for each detected instance.

InstanceMaskObservation

A dedicated VNInstanceMaskObservation wrapper.

SegmentationMask

A single grayscale mask in row-major byte order.

Enums

SegmentationQuality

Apple person-segmentation quality.

Functions

generate_foreground_instance_mask_in_path

Generate an instance segmentation mask of all foreground objects in the image at path (macOS 14+).

generate_foreground_instance_mask_observation_in_path

Generate a dedicated VNInstanceMaskObservation wrapper for the image at path.

generate_person_segmentation_in_path

Generate a person/body silhouette mask for the image at path.