How Zurich Is Tackling the Duplicate Image Problem in City Records — and Where It Stands Against Rivals

Zurich's Stadtarchiv and the cantonal building registry together hold an estimated 14 million digitised image files accumulated since the early 2000s — and a growing share of those files are exact or near-exact duplicates that slow searches, inflate storage costs, and create legal ambiguity when planners pull construction records. The city's IT department quietly began a structured deduplication programme in the first quarter of 2026, targeting the Bau- und Wohnungsdirektion's permit archives first.

The timing is not coincidental. A federal directive issued by the Swiss Federal Archives in Bern in late 2025 called on cantonal and municipal bodies to demonstrate compliance with updated records-management standards by the end of 2026. Duplicate image files sit at the centre of those standards: under Swiss archival law, a document with multiple identical digital representations must be rationalised to a single authoritative copy, or each version must carry explicit metadata explaining why it exists. For a city the size of Zurich — processing roughly 9,200 building permits per year according to figures published by Stadtentwicklung Zürich — that is a considerable administrative burden.

What Zurich Is Actually Doing

The practical work is happening at two nodes. The Amt für Städtebau, headquartered on Lindenhofstrasse, began piloting perceptual-hash deduplication software in February 2026, scanning approximately 400,000 permit-related images. The tool flags pairs with more than 98 percent pixel similarity, then routes them to a human reviewer before any deletion is confirmed. Errors at this stage have downstream consequences: a wrongly merged planning image contributed to a disputed variance case in Wiedikon in 2023, a cautionary example officials cite internally when justifying the manual review step.

Separately, the Stadtarchiv on Neumarkt is running a parallel project for historical photographic collections, working with ETH Zürich's Computer Vision Lab to classify near-duplicate images of city infrastructure taken across different decades. That collaboration, formalised in a memorandum of understanding signed in March 2026, is designed to produce open-source tooling that smaller Swiss municipalities — St. Gallen, Winterthur, Biel — can adopt without procuring expensive commercial licences.

The cost picture matters. Commercial deduplication platforms aimed at municipal archives carry licensing fees that typically start at CHF 40,000 per year for a deployment of Zurich's scale, according to procurement documents in the public tender register. The ETH collaboration model aims to bring that recurring cost closer to zero for the open-source tier, though infrastructure and personnel costs remain.

How Zurich Compares Internationally

Other European cities are grappling with the same problem at varying stages of seriousness. Amsterdam's Stadsarchief disclosed in its 2025 annual report that roughly 11 percent of its digitised collection contained duplicate or near-duplicate entries — a figure that surprised city councillors and prompted an emergency remediation budget of €620,000. Vienna's Wiener Stadt- und Landesarchiv began a similar audit in 2024 but has yet to publish results. Copenhagen's city archive, working with the Danish National Archives, completed a deduplication sweep in late 2024 and reported a 17 percent reduction in total stored image volume, freeing meaningful server capacity in its Ørestad data centre.

Zurich's approach is more conservative than Copenhagen's in one important respect: the city has not set a hard deletion quota. Every flagged duplicate triggers a review rather than an automatic purge. That caution reflects both Swiss legal culture and the city's experience with the Wiedikon case, but it also means the programme moves more slowly. Staff at the Amt für Städtebau estimate full processing of the permit archive will take until mid-2027.

For residents and professionals who regularly pull planning records — architects working on Langstrasse renovations, developers tracking density rules in Altstetten — the practical payoff should be faster search response times and fewer instances of conflicting image versions appearing in the same record bundle. The city's online building permit portal, last redesigned in 2022, is expected to reflect the cleaned archive in a phased rollout beginning in the first quarter of 2027. Until then, users encountering duplicate images in a permit file can flag them through the existing feedback form on the Stadtarchiv website, where submissions are reviewed weekly.