Conference Agenda

Research in machine learning increasingly depends on image datasets that undergo iterative processes of annotation, inspection, correction, and reuse. However, current data publishing practices remain largely file-centric: datasets are released as archives or object storage buckets with limited support for preview, structured inspection, or transparent documentation of annotation provenance. This creates friction for reuse, peer review, and reproducibility, particularly as datasets evolve over time.

This presentation explores the potential of the IIIF Image and Presentation APIs together with the W3C Web Annotation Data Model as a representation and delivery layer for research image datasets outside of cultural heritage. The exploration is grounded in a real-world use case: post-consumer plastic sorting for recycling, where image datasets support tasks such as identifying recyclable items, separating valuable objects, and detecting contaminants. These datasets are visually complex, affected by occlusion and ambiguity, and subject to iterative relabeling driven by expert review and model feedback.

Rather than presenting a finalized system, this work takes an exploratory approach. Building on existing data pipelines, it examines where and how IIIF services and Web Annotation concepts can be integrated to improve dataset inspectability, transparency, and long-term reusability.

In 2019, the National Museum of African American History and Culture (NMAAHC) and Getty jointly acquired the photo archive of Johnson Publications, the publisher of Ebony, Jet, and 12 other magazines that documented American culture through a Black perspective. This archive, consisting of over 4 million photos, serials, and audio-visual materials, is being completely digitized and made available to the public as a joint project of the two organizations.

Andrew Gunther, Lead Data + Imaging Egress Engineer at Smithsonian and David Newbury, Senior Director, Public Technologies at the Getty will discuss how IIIF is the key enabling technology of this collaboration, where digitization and digital asset management are being provided within the Smithsonian infrastructure and metadata management and access is being provided via the Getty infrastructure. We will discuss both the business and technical complexities of this sort of collaboration, sharing how IIIF is used to bridge between our two infrastructures.

What happens to a project website when the project funding ends? Is it closed, or does the institution attempt to maintain it?

In the Notke research project, IIIF was integrated from the very beginning as both a data delivery mechanism and a core design principle. The project content was structured so that users can achieve a comparable research experience through an IIIF viewer to that offered by the original bespoke website. The project website brings together high-resolution images, 3D models, and data derived from a range of instrumental analyses. Rather than relying solely on a custom-built web application, the research outputs are structured around a simple IIIF manifest and delivered through an institutional asset management system and image server. This approach ensures that the core research materials remain accessible, reusable, and sustainable beyond the lifespan of the original website.

By embedding IIIF as the primary method of data delivery, the project demonstrates how research outputs can be preserved and made available even after funding ends or when there are no longer resources to maintain a dedicated web application. This model supports long-term institutional sustainability while improving accessibility and reuse of digital humanities research.

The creation of 3D-models of most important archaeological findings from Styria - Roman temple architecture, sculptures and inscriptions, was the focus of the Austrian project Digital Temple in 2025. The scientific aim of this project was to analyse the architectural elements and develop a reconstruction of one of the region's most important sanctuaries. In addition, the objects were to be made available in high-resolution 3D models for scientists, museum visitors and users of the European Data Space for Cultural Heritage (EDSCH).

In 2026, the follow-up project DOMI will introduce a workflow that supports the automated transformation of the descriptive data of 3D models into the IIIF standard using IIIF Presentation API v4. Access to the published 3D data will be diverse, via museum homepages, national Austrian cultural heritage portals and the EDSCH. In various use cases, selected data will be enriched with annotations, datasets will be published for further processing with artificial intelligence, and an IIIF Simple Collection will be compiled. Above all, a sustainable workflow will be established with customised tools for local museum staff, taking into account international standards (IIIF, W3C annotation) and latest documentation guidelines (EDM for 3D, data sheets for datasets).