Does it include documentation for Apple's 'Subject Lifting'?

Absolutely. It covers both VisionKit's ImageAnalysisInteraction for UI-based lifting and the programmatic VNGenerateForegroundInstanceMaskRequest for custom implementations.

Does this skill support the latest visionOS?

Yes, the axiom-vision-ref skill includes API references for visionOS 1+ and the most recent Vision framework updates through iOS 26+.

Can I use this for real-time video processing?

Yes, the skill provides patterns for using VNImageRequestHandler and includes performance tips for running requests on background queues to avoid blocking the UI.

How do I handle the Vision coordinate system?

The reference explains that Vision uses a normalized coordinate system (0.0-1.0) with a lower-left origin, which is different from standard UIKit coordinates.

Axiom Vision Framework Reference

Name: Axiom Vision Framework Reference
Author: CharlesWiltgen

byCharlesWiltgen

•

233

•

Mobile Development

Provides comprehensive API documentation and implementation patterns for Apple's Vision framework across xOS platforms.

This skill serves as a specialized technical reference for integrating Apple's Vision framework into iOS, iPadOS, macOS, and visionOS applications. It offers detailed guidance on high-level computer vision tasks such as subject segmentation (subject lifting), 2D and 3D hand and body pose detection, face analysis, and OCR. By providing precise API signatures, coordinate system mappings, and performance best practices, it helps developers implement advanced visual features like gesture recognition, background removal, and document scanning within their Claude Code workflow.

主要功能

01Detailed 21-point hand and 18-point body pose landmark mapping

02233 GitHub stars

03Reference for OCR and barcode detection using VNRecognizeTextRequest

04Multi-person segmentation and hit-testing logic for pixel buffers

05Latest xOS compatibility including 3D body pose and visionOS support

06Implementation guides for VisionKit 'Subject Lifting' and instance masking

使用场景

01Building high-performance document scanners and OCR data extractors

02Implementing gesture-based controls using real-time hand pose tracking

03Creating photo editing tools with automatic subject extraction and segmentation

主要功能

01Detailed 21-point hand and 18-point body pose landmark mapping

02233 GitHub stars

03Reference for OCR and barcode detection using VNRecognizeTextRequest

04Multi-person segmentation and hit-testing logic for pixel buffers

05Latest xOS compatibility including 3D body pose and visionOS support

06Implementation guides for VisionKit 'Subject Lifting' and instance masking

使用场景

01Building high-performance document scanners and OCR data extractors

02Implementing gesture-based controls using real-time hand pose tracking

03Creating photo editing tools with automatic subject extraction and segmentation