michina-swift

KO | EN

gitlite — search

#hacktoberfest #react #ai #llm #golang #android #deep-learning #macos #java #windows #cli #awesome-list #vue #mcp

(★ 9)

Immich Machine Learning Server on macOS.

michina-swift 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Michina

Immich Machine Learning Server on macOS.

[![Build Release][workflow-build-release-badge]][workflow-build-release]
[![TestFlight](https://img.shields.io/badge/TestFlight-join-blue)](https://testflight.apple.com/join/rJP9acvk)

[workflow-build-release]: https://github.com/lucka-me/michina-swift/actions/workflows/build-release.yml
[workflow-build-release-badge]: https://github.com/lucka-me/michina-swift/actions/workflows/build-release.yml/badge.svg

> [!IMPORTANT]
> This project is still under development. **Use AT YOUR OWN RISK, and do not forget to BACKUP your database before
> running massive machine learning jobs on your Immich instance with Michina.**

## Glance

In a nutshell, Michina is a Swift implementation of
[immich/machine_learning](https://github.com/immich-app/immich/tree/main/machine-learning), with a GUI, for macOS.

> [!NOTE]
> Because of the differences in image processing ecosystem between of Swift (Core Image, Core Graphic, vImage...) and
> Python (Pillow, OpenCV, NumPy...), Machina doesn't always behave exactly the same as the original official
> implementation does, which may lead to slight differences in the outputs.  
> But these differences are expected to be toleratable, like the facial recognition embedding being different but still
> close enough to be classified as the same person.

## Packages

This Xcode Workspace contains following local packages, READMEs are available.

- [Magearna](./Magearna): Core component of Michina, provides basic inference features.
- [ONNXRuntime](./ONNXRuntime): The fundamental library to run models.
- [Tokenizers](./Tokenizers): Tokenize text for Textual Smart Search, bridged from Rust.

## Build

Some XCFramework and source file of the packages need to be built or generated before resolving, please refer to their
`README.md` for details of building.

## Execution Providers

Backed by ONNXRuntime, Michina can run models with Core ML framework or directly with CPU, please refer to the
[benchmark](./docs/content/benckmarks.md) ([web page](https://michina.lucka.dev/benckmarks)) for their comparison.

michina-swift

michina-swift 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

michina-swift

michina-swift 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README