PictoPy: Expanding into Semantic Intelligence, Memories, Video, and Secure Sharing

This proposal aims to implement intelligent event classification, natural language search, and EXIF metadata read-write capabilities for image metadata portability. Key backend upgrades include implementing image metadata based image clustering with SigLIP2 for visual similarity-driven clustering. It introduces application optimization techniques like modular AI model management, allowing users to extend functionality on demand without bloating the base install, to keep the core application lightweight. It improves face clustering accuracy by addressing confirmed gaps in the current DBSCAN pipeline, including adaptive eps estimation and a face quality gate. On the frontend side, a new Memories UI will be introduced with advanced sorting features, user notifications, video support, along with a new landing page for the product launch. A major addition is shareable album links: users will be able to generate secure, token-based links to share selected photos over the network while the PictoPy backend is running, with new API endpoints serving image data directly to a lightweight web-based receiver. Additionally, comprehensive documentation—including contributing guides, API references, and setup videos—will encourage new developers to contribute. These improvements will make PictoPy more user-friendly, scalable, and ready for widespread adoption.