Does this support the latest Apple platform features?

Yes, it includes support for newer features like 3D Body Pose (iOS 17+), Person Instance Segmentation, and the new RecognizeDocumentsRequest available in recent OS versions.

How do I ensure Vision processing doesn't lag my app?

Never run Vision requests on the main thread. This skill teaches you how to set up dedicated background processing queues and optimize camera frame rates to keep the UI responsive.

Can I isolate an object without including the person's hand holding it?

Yes, this skill provides a specific pattern that combines subject segmentation with hand pose detection to create an exclusion mask, effectively removing the hand from the object's bounding box.

Should I use VisionKit or the Vision framework?

Use VisionKit when you need system-standard UI behaviors (like long-press subject lifting). Use the Vision framework for custom processing pipelines, working with HDR images, or low-level tasks like pose detection.

Axiom Vision for xOS

Name: Axiom Vision for xOS
Author: CharlesWiltgen

byCharlesWiltgen

•

148

Mobile Development

Implement advanced computer vision features including subject segmentation, pose detection, and OCR for Apple platforms.

概要

The Vision skill provides Apple developers with a comprehensive toolkit for integrating sophisticated computer vision capabilities into iOS, macOS, visionOS, and tvOS applications. It offers specialized implementation patterns for subject lifting, hand and body pose tracking, text recognition, and barcode scanning while navigating the nuances between the Vision and VisionKit frameworks. By providing clear decision trees and performance-focused code examples, it helps developers solve complex problems like isolating objects from hands or preserving HDR during compositing without blocking the main UI thread.

主な機能

148 GitHub stars
Subject segmentation and instance-based 'subject lifting' from backgrounds.
Text recognition (OCR) and barcode/QR code scanning with DataScannerViewController.
Architecture guidance for choosing between VisionKit and raw Vision APIs.
Real-time hand and body pose detection for gestures and action classification.
Document scanning with automated perspective correction and data extraction.

ユースケース

Creating photo editing apps that isolate subjects for custom backgrounds while preserving HDR.
Building fitness or health apps that track real-time body movement and 3D pose landmarks.
Developing high-performance scanning tools for retail or logistics using OCR and barcode detection.

概要

主な機能

148 GitHub stars
Subject segmentation and instance-based 'subject lifting' from backgrounds.
Text recognition (OCR) and barcode/QR code scanning with DataScannerViewController.
Architecture guidance for choosing between VisionKit and raw Vision APIs.
Real-time hand and body pose detection for gestures and action classification.
Document scanning with automated perspective correction and data extraction.

ユースケース

Creating photo editing apps that isolate subjects for custom backgrounds while preserving HDR.
Building fitness or health apps that track real-time body movement and 3D pose landmarks.
Developing high-performance scanning tools for retail or logistics using OCR and barcode detection.