This blueprint helps you build your own local image organization and search system using open-source tools. It leverages image-to-text and text embeddings models to generate image descriptions, create embeddings, and store them in a vector database for efficient retrieval. Follow this Blueprint to organize, search, and categorize your image collection—all locally on your machine. For the image-to-text model, we recommend Mozilla’s distilvit model, due to its small size and debiased training methodology.
