TagGUI is a cross-platform desktop application designed for efficient management and captioning of image datasets, primarily targeting users creating data for generative AI models like Stable Diffusion. It offers a keyboard-friendly interface, tag autocompletion, integrated Stable Diffusion token counting, and automatic caption/tag generation using various models.

How It Works

TagGUI operates as a desktop application, allowing users to load directories of images. Tags are managed via associated .txt files, with changes automatically saved. A key feature is its auto-captioning capability, leveraging models like CogVLM, LLaVA, and WD Tagger. Users can select images, choose a captioning model, and provide prompts with template variables for dynamic content insertion. GPU acceleration is supported for NVIDIA hardware, with CPU fallback available.

Quick Start & Requirements

• Installation: Download the latest release executable for your OS from the releases page. Extract and run.
• Prerequisites:
  • Python 3.12 (recommended) or 3.11.
  • macOS users may need manual installation.
  • Linux users might require libxcb-cursor0 and potentially python3.12-dev or python3.11-dev for CogVLM2.
  • NVIDIA GPU with CUDA support is required for GPU-accelerated captioning.
• Setup: Download and extract. Manual installation involves cloning the repo and installing from requirements.txt.
• Links: Releases page: https://github.com/jhc13/taggui/releases

Highlighted Details

• Supports automatic caption/tag generation with models including CogVLM, LLaVA, and WD Tagger.
• Features advanced image list filtering with support for tag, caption, name, path, character count, and token count criteria, combinable with logical operators (AND, OR, NOT) and wildcards.
• Keyboard-centric interface for rapid tagging, including tag autocompletion and batch operations.
• Integrated Stable Diffusion token counter.

Maintenance & Community

• The project appears to be maintained by a single developer, jhc13.
• No explicit community links (Discord, Slack) or roadmap are provided in the README.

Licensing & Compatibility

• The README does not explicitly state a license. The repository's license file should be consulted for details.

Limitations & Caveats

• No official macOS release is provided due to lack of testing hardware.
• Some advanced model features might require specific Python development headers on Linux.
• The README does not mention specific performance benchmarks or detailed hardware requirements for auto-captioning beyond GPU necessity.