dallaccumulo

Dallaccumulo is an open‑source software framework that combines the image synthesis capabilities of OpenAI’s DALL‑E generative models with the distributed, scalable storage and retrieval features of Apache Accumulo. The project was initiated in 2022 by a research team at the University of Data Science to address the need for efficient management of large collections of AI‑generated imagery in scientific and creative research workflows. The team released the first version through GitHub under a BSD‑3 clause license, encouraging contributions from developers in both the AI and big data communities.

The core architecture of Dallaccumulo comprises two primary modules. The front‑end module interfaces with the DALL‑E API to generate images based on textual prompts, then embeds metadata such as prompt, generation timestamp, and model parameters. The back‑end module writes the resulting image blobs and metadata to an Accumulo instance, using its key‑value store to support fast, fine‑grained queries. Clients can retrieve images by keyword, date range, or embedding similarity through Accumulo’s iterator API. The framework also includes a RESTful service layer that simplifies interactions for web and command‑line clients.

Typical use cases involve researchers creating annotated datasets for computer vision, designers prototyping product concepts, or archivists storing and cataloguing generative art collections. By leveraging Accumulo’s strong consistency guarantees and capabilities for handling very large datasets, Dallaccumulo supports workflows that handle millions of images without compromising retrieval speed. The project is maintained by a core team of five developers and relies on a growing ecosystem of contributors, with regular code reviews and automated testing integrated into its release pipeline.