Zurich Pushes Forward on Duplicate Image Purge: What Happened This Week

The problem has been building for years, but this week a cluster of developments put duplicate image management squarely on Zurich's digital infrastructure agenda. ETH Zurich's Media Technology Center, based on the Hönggerberg campus, confirmed it had completed the first phase of an automated deduplication audit covering more than 1.2 million image files stored across its internal research repositories. The audit, which ran through the end of June 2026, identified roughly 340,000 redundant files — nearly 28 percent of the total archive — flagged for review or deletion.

The timing is not accidental. Across Zurich's public institutions, the pressure to rationalise digital storage has grown as cloud hosting costs have risen and data governance requirements under Switzerland's revised Federal Act on Data Protection, which came into force in September 2023, have tightened. Holding duplicate images of identifiable people without documented purpose now carries genuine legal exposure, not just administrative inconvenience.

Archives, City Hall, and a Shared Problem

The City of Zurich's Stadtarchiv, located on Neumarkt in the Altstadt, has been running its own parallel effort. Staff there are working through a digitised collection of historical press photographs, some dating to the 1940s, where multiple scans of the same print were made at different resolutions over successive digitisation projects. The archive's current consolidation programme, launched in spring 2025, is scheduled to conclude by the end of the third quarter of 2026. Duplicate handling is one of three workstreams, alongside metadata standardisation and rights clearance.

NZZ Mediengruppe, whose editorial offices sit on Falkenstrasse in Zurich's 1st district, has been dealing with the same issue at a commercial scale. The publisher has not disclosed figures publicly, but industry observers note that large Swiss news organisations typically maintain image libraries running into the tens of millions of files accumulated over decades of digital publishing, with duplication rates in legacy systems often exceeding 20 percent before active management begins.

The practical trigger this week was a software update. Axiell, the Swedish collections management vendor used by several Swiss cultural institutions, pushed a new deduplication module to its AMU platform on July 1. The Zurich Central Library — the Zentralbibliothek on Zähringerplatz — is among the Swiss institutions running Axiell systems, and library technology staff confirmed they are evaluating the new module for a trial rollout later this summer. The module uses perceptual hashing, a technique that identifies visually near-identical images even when file names or metadata differ, and flags them for human review rather than deleting automatically.

Why Perceptual Hashing Changes the Calculus

Earlier deduplication tools relied on exact file matching — same pixel dimensions, same file size, same checksum. Those tools missed the vast category of near-duplicates: the same photograph cropped differently, resaved at a different compression level, or lightly colour-corrected for different print and web uses. Perceptual hashing closes that gap, which is why institutions that ran exact-match deduplication five years ago are now finding they still have significant redundancy.

For Zurich's housing-stretched public sector, there is also a cost argument. Municipal IT budgets across Switzerland have faced pressure as the broader Wohnungsnot crisis has pushed capital expenditure toward social infrastructure rather than back-office systems. Reducing storage bloat is one of the few areas where IT departments can show direct savings without new spending — cloud storage pricing typically falls in the range of CHF 0.02 to CHF 0.04 per gigabyte per month for institutional contracts, and a reduction of hundreds of thousands of files adds up across a multi-year contract horizon.

For organisations in Zurich still working through their own duplicate image backlogs, the practical advice from digital archivists is consistent: do not delete automatically. The current best practice is a three-stage workflow — flag, human review, then archive or delete — with deletion logs retained for at least five years to satisfy audit requirements under Swiss law. The Zentralbibliothek's planned trial this summer will be watched closely by peer institutions across the city, and results are expected to be presented at a joint archive and library working group meeting in Bern in October 2026.