This repository is a curated list of machine learning, AI, and NLP resources specifically for iOS developers, addressing the challenge of finding suitable tools written in iOS-friendly languages. It provides a comprehensive overview of libraries, frameworks, APIs, and learning materials for on-device ML and related fields.

How It Works

The list categorizes resources by ML domain (Core ML, Computer Vision, NLP, etc.) and implementation language (Swift, Objective-C, C++, JavaScript). It highlights libraries that can be easily integrated into iOS applications, including those that leverage Apple's Core ML framework and Metal Performance Shaders for hardware acceleration. The compilation aims to bridge the gap between powerful but often platform-unfriendly ML tools (Python, Java) and the needs of mobile development.

Quick Start & Requirements

• Installation: Primarily involves integrating libraries via CocoaPods, Swift Package Manager, or direct code inclusion.
• Prerequisites: Swift, Objective-C, C++, JavaScript development environments. Some libraries may require specific iOS SDK versions or Metal support.
• Resources: Links to official documentation, GitHub repositories, and tutorials are provided for most entries.

Highlighted Details

• Extensive coverage of Core ML, including converters and pre-trained models.
• Numerous Swift and Objective-C libraries for general ML, deep learning, and specific tasks like NLP and computer vision.
• Inclusion of low-level routines like Apple's BNNS and MetalPerformanceShaders for optimized on-device inference.
• Categorization of Web APIs for cloud-based ML services.

Maintenance & Community

• The project is a personal curation, last updated January 12, 2018.
• Pull requests are welcome for contributions.

Licensing & Compatibility

• Licenses vary widely across the listed resources, including Apache 2.0, MIT, BSD, GNU GPL, and LGPL.
• Compatibility for commercial use depends on the specific license of each included library.

Limitations & Caveats

The repository's last update was in early 2018, meaning many of the listed resources may be outdated or superseded by newer technologies and frameworks. The "work in progress" status is noted for some entries.