Paris Officials and Heritage Experts Demand Crackdown on Duplicate Images Flooding City's Cultural Archives

The BnF has confirmed it is sitting on a problem it can no longer ignore. Duplicate and mislabelled photographic records now account for an estimated 12 to 18 percent of the institution's Gallica digital archive — roughly 400,000 redundant image files, by internal assessments shared with city council members last month — creating cataloguing chaos that is delaying research projects and distorting automated search results across public databases.

The timing is pointed. Paris is deep into activating its post-2024 Olympics legacy infrastructure, a process that depends heavily on accurate digital documentation of urban change along the Seine corridor and in the banlieues served by the Grand Paris Express. When image records are duplicated, misfiled or misattributed, planners lose the baseline visual evidence they need to measure transformation. A street in Saint-Denis photographed in 2019 ends up filed twice under different metadata, one version tagged to Aubervilliers, and suddenly a housing impact assessment is built on a cracked foundation.

"The catalogue is not decorative — it is a working tool," one senior archivist at the Musée Carnavalet, on Rue de Sévigné in the Marais, told colleagues at a professional seminar held in late June. The museum, which reopened after a four-year renovation in 2022, has been digitising its collection of over 600,000 Paris-specific images, and staff say duplicate detection algorithms currently flag inconsistencies but lack the authority protocols to auto-resolve or delete without human sign-off — a bottleneck producing a backlog of more than 50,000 unresolved cases as of June 2026.

What the Experts Are Recommending

Specialists in digital heritage management are pushing for a shared deduplication framework across municipal institutions. The Paris Musées network, which coordinates 14 city-owned museums including the Petit Palais on Avenue Winston Churchill and the Maison de Victor Hugo on Place des Vosges, has been floating a proposal since April for a centralised image identity layer — essentially a shared fingerprinting system that would flag identical or near-identical files before they are ingested into any member institution's database.

The Institut national du patrimoine, which trains France's conservateurs and archivistes, held a working group on the issue in May at its premises near the Bourse de Commerce. Participants from the Atelier parisien d'urbanisme, known as APUR, stressed that the problem is not merely administrative. APUR publishes visual documentation of housing stock, public space and demographic change across Paris's 20 arrondissements and the inner suburbs, and its analysts say duplicate images are beginning to skew the machine-learning models the agency uses to detect illegal subdivisions and monitor rental market pressure in areas like the 18th arrondissement and Pantin.

Data from the Direction régionale des affaires culturelles Île-de-France, released in May, showed that public institutions in the region spent approximately €2.3 million on digitisation contracts in 2025 alone. Critics argue a meaningful portion of that budget is effectively wasted when the resulting files are never properly deduplicated or cross-referenced between institutions that share subjects — particularly photographs of the Seine banks, Haussmann facades, and infrastructure sites now under active Grand Paris Express construction.

Pressure on the Mairie to Act

Inside the Hôtel de Ville, there is growing pressure on the cultural services directorate to formalise a deduplication standard before the end of 2026. City councillors from the list aligned with Mayor Anne Hidalgo have asked for a progress report on the Paris Musées digitisation charter by September. A deputy mayor's office confirmed the issue is on the agenda for autumn budget negotiations, though no specific funding line has been announced.

For researchers and urban planners, the practical advice from experts right now is blunt: audit your own institution's holdings before submitting image batches to any shared platform. The BnF has posted updated ingest guidelines on its professional portal, requiring submitting partners to run at least one automated hash-comparison check before transfer. Institutions that skip that step will, starting in January 2027, face a mandatory quarantine period during which their submissions will not appear in Gallica's public-facing search. The deadline is firm, archivists say. The backlog is not getting smaller on its own.