I've just vibecoded a replacement for tagGUI (as it's abandoned)

Developer artemyvo has launched ImageTagger on GitHub, creating a functional replacement for the discontinued tagGUI image tagging software. Built as an open-source Python application, it restores core functionality for managing tags across image libraries. The project's initial release includes basic tagging operations, providing a foundation for users who relied on the original tool.

The standout feature is seamless integration with Ollama, the popular framework for running local large language models. By connecting ImageTagger to vision-capable models like LLaVA or Claude 3.5 Sonnet via Ollama, users can automate tag generation directly from image content. Early testing shows this AI-assisted approach not only creates relevant tags but also validates existing metadata, helping identify inconsistencies and errors in large datasets. This functionality addresses a critical pain point in machine learning workflows where clean, well-labeled training data is essential for model performance.

The tool represents a practical application of multimodal AI for data management tasks. By leveraging locally-run vision models, ImageTagger offers privacy-preserving automation without relying on cloud APIs. The developer reports that initial validation tests on existing libraries have produced 'interesting insights' for dataset cleaning, suggesting potential time savings for photographers, researchers, and ML practitioners managing visual assets. As an open-source project, it invites community contributions to expand its capabilities beyond the restored tagGUI feature set.