How Melbourne's Public Art Collection Ended Up With Thousands of Duplicate Digital Images — and What Comes Next

Melbourne's major cultural institutions are sitting on a quiet administrative crisis. Across the City of Melbourne's art collection, the State Library Victoria on Swanston Street, and the Melbourne Museum in Carlton, curators and digital archivists have spent much of 2025 and early 2026 confronting the same stubborn problem: thousands of duplicate image files embedded inside public-facing collection databases, many of them linked to the wrong catalogue records entirely.

The issue matters now because the Victorian Government's Creative State 2025–2028 strategy sets explicit benchmarks for digitally accessible public collections. Institutions under that framework are expected to have clean, searchable, non-redundant digital records by mid-2027. With 13 months left on that clock, the scramble to audit and replace duplicate entries has become one of the less glamorous — but increasingly urgent — pieces of arts infrastructure work in the state.

How the Duplicates Accumulated

The problem has a mundane origin story. Through the 2010s, galleries and libraries migrated legacy card-catalogue records into new content management systems at different times, using different metadata standards. The State Library Victoria completed a major platform migration in 2016. The Melbourne Museum, managed by Museums Victoria, ran a parallel digitisation push around 2018 and 2019. The City of Melbourne's art collection — which spans works displayed in council buildings from the Town Hall on Swanston Street to offices in Docklands — went through its own piecemeal digitisation process without a unified naming convention for image files.

Each migration imported earlier errors. A photograph scanned twice under slightly different file names, or a high-resolution image attached to the wrong accession number, would propagate through backup systems and public portals. Staff turnover compounded the problem; institutional memory of which file was the authoritative version eroded over time. By early 2025, internal audits at more than one of these institutions had flagged that some individual artworks had between three and seven separate image files sitting in live databases, not all of them identical in resolution or cropping.

Museums Victoria, which oversees the Melbourne Museum, has publicly acknowledged the scale of its digitisation program in annual reports. Its collection holds more than 17 million objects, though only a fraction are accompanied by digital images. The sheer volume means even a small percentage of duplicate or misattributed image records represents tens of thousands of individual files requiring human review.

The Remediation Process

The current remediation approach involves a combination of automated deduplication software and manual curatorial checking. Automated tools can flag probable duplicates based on file-hash matching and image-similarity algorithms, but they cannot make the final call on which version to keep — that still requires a trained eye to assess image quality, cropping accuracy, and correct metadata linkage.

The City of Melbourne began a dedicated image-replacement audit in late 2024, prioritising the roughly 4,200 works in its public art collection that are accessible through the Melbourne Art Tram program and the online Art and Heritage Collection portal. According to the council's own published collection strategy, the portal went live in its current form in 2021, meaning some duplicate records have been publicly visible for close to five years.

The work is slow. An archivist reviewing disputed image records at a major state institution can typically resolve between 80 and 120 records per day, depending on the complexity of provenance documentation involved. At that rate, a backlog of 10,000 flagged files represents roughly four to five months of dedicated labour for a single full-time specialist.

For researchers and educators who use these portals — school teachers pulling images for curriculum work, academics at the University of Melbourne's Art History department on Parkville campus, or independent curators preparing touring exhibitions — the practical impact has been patchy search results and occasional mislabelled reproductions circulating in published materials.

The immediate priority for institutions under the Creative State framework is to complete first-pass deduplication before the end of 2026, leaving the first half of 2027 for quality assurance and public portal updates. Researchers relying on any of Melbourne's major online collection portals for publication work would be well advised to cross-reference image metadata against physical catalogue records held at each institution, at least until the audits are formally signed off.