Curator is a self-hosted application for managing image collections with advanced search, tagging, and organization capabilities.
Curator is designed to help you organize, search, and browse your image collections with powerful features:
- Vector-based similarity search: Find visually similar images using CLIP neural embeddings
- Intelligent deduplication: Prevent duplicate uploads through content-based image hashing
- Advanced tagging system: Organize images with a flexible tagging system
- People management: Track creators and subjects in your images
- Metadata enrichment: Store titles, descriptions, and source information
This is a work-in-progress prototype, and many critical features are still missing or incomplete. The current codebase prioritizes functionality over perfect code quality as we validate core concepts.
- No authentication or multi-user support
- No user interface at this stage
- Missing features like collections and tag hierarchies
- Code needs refactoring and optimization
- No test coverage
- Backend: Go with Echo framework
- Database: PostgreSQL with pgvector extension
- Search: Elasticsearch for metadata, Qdrant for vector similarity
- ML: OpenAI CLIP for generating image embeddings
As this is an early prototype, we're not actively seeking contributions yet, but feel free to open issues for feature suggestions or bug reports.