# LLMII (ImageIndexer)

> Self-hosted desktop app using a local vision LLM to caption and keyword images, embedding results in standard metadata fields for any photo manager.

- **Category:** Photo Management
- **Licence:** GPL-3.0
- **Open source:** oss
- **Languages:** Python
- **Platforms:** Linux, Windows, macOS
- **Tags:** Cross Platform, Metadata, Tagging, Tags, Multi modal, Image Processing, AI, Self Hosted, Photo Management, Keywords, Local, Llama.cpp, Koboldcpp, Local llm, Image recognition, Large language models, Image Classification, Exiftool, Vision Language Model, Dataset Generation, Exif Metadata

## Description

LLMII is a cross-platform desktop tool for photographers who want AI-powered image tagging without cloud uploads. It bundles KoboldCPP as a local inference backend, downloads a vision-language model on first run, and uses ExifTool to embed generated captions and keywords into each image's metadata or an XMP sidecar. The GUI lets you choose one-step or two-step processing, adjust sampler settings, skip folders, reprocess failures, and review results in an image viewer. All processing happens locally on Windows, macOS, or Linux; a discrete GPU with 8 GB VRAM is recommended for usable speed. Licensed GPL-3.0 with no paid tiers.

## Links

- [Homepage](https://github.com/jabberjabberjabber/imageindexer)
- [GitHub Repository](https://github.com/jabberjabberjabber/ImageIndexer)
- [Wiki / Documentation](https://github.com/jabberjabberjabber/ImageIndexer/wiki)

---
Source: https://selfhost.directory/project/imageindexer