Prague's Digital Archives Are Drowning in Duplicate Images — and the Numbers Tell a Sobering Story

Prague's municipal digitisation program is sitting on a problem that nobody talks about at press conferences: duplicate images. An internal review conducted across several of the city's public-facing digital repositories in the first half of 2026 found that between 18 and 22 percent of all stored image files are exact or near-exact copies of files already held elsewhere in the same system. For a city that has poured hundreds of millions of crowns into digital infrastructure over the past decade, that is a significant slice of wasted capacity.

The issue matters now because Prague City Hall is midway through expanding its open-data portal, data.Praha.eu, with the aim of making urban planning documents, heritage photography and public event records fully searchable by the end of 2027. Storage redundancy at this scale does not just inflate server costs — it undermines search accuracy, slows retrieval times and creates version-control headaches for archivists trying to maintain a clean public record.

Where the Problem Is Concentrated

The duplication burden is not spread evenly. The heaviest concentrations appear in collections tied to Prague's heritage districts. The Institute of Planning and Development — whose offices sit on Vyšehradská street in Nusle — manages photographic records covering construction permits and façade approvals across the city's protected zones, including Malá Strana and Hradčany. Staff there have flagged that automated batch-scanning processes used between 2019 and 2023 routinely produced two or three copies of each image without a consistent deduplication step before upload.

The Prague City Archives on Archivní street in Holešovice faces a related challenge with its digitised collection of historical photographs, some dating to the late nineteenth century. Because different departments commissioned separate scanning projects under separate contracts, the same glass-plate negative has in some cases been digitised four times and stored across four different folders. Archivists working on the project — which falls under the broader Smart Prague 2030 strategy — describe the reconciliation work as painstaking and under-resourced.

What the Data Actually Shows

Figures compiled by the city's IT department and shared with councillors on the Committee for Digitalisation in May 2026 put the total volume of image files across municipal systems at roughly 4.7 million files. Of those, an estimated 900,000 are flagged as probable duplicates — either pixel-identical copies or images differing only in compression level or file format. At current cloud-storage contract rates paid by the city, maintaining those redundant files costs an estimated 1.2 million crowns per year in unnecessary expenditure, according to the committee briefing document.

The replacement or deduplication of those files is not a trivial operation. A pilot run carried out in January 2026 on a subset of 50,000 images from the transport planning database — covering road works records in districts Praha 4 and Praha 9 — took a three-person technical team six weeks to complete. That pace, if applied to the full 900,000-file problem, implies a multi-year effort without additional staffing or automated tooling.

Procurement documents published on the city's official tender portal in June 2026 show that Prague is now seeking a supplier for deduplication software capable of processing at least 200,000 image files per week with a false-positive rate below 0.5 percent. The contract value listed in the notice is up to 3.8 million crowns. Bids were due by 30 June, and the city expects to announce a preferred vendor before the summer recess ends in late August.

For residents and researchers who rely on the open-data portal — including journalists, property developers pulling planning records and academics studying historic Žižkov or Vinohrady — the practical upshot is straightforward: search results will remain noisy and occasionally contradictory until the deduplication work is done. The city's stated target is to reduce duplicate image prevalence below 3 percent across all core repositories by the end of 2028. Whether the budget and timeline hold will depend heavily on which vendor wins the June tender and how quickly an implementation contract can be signed.