LA's War on Duplicate Images: How the City Stacks Up Against London, Tokyo and São Paulo

Los Angeles city archivists quietly flagged a problem last fall that had been festering for years: the Bureau of Engineering's public-facing digital asset library had accumulated more than 340,000 duplicate image files — redundant photographs of streetscapes, infrastructure sites and housing developments that were eating storage, slowing database queries and, in several cases, sending permit reviewers to look at the wrong version of the same block. The clean-up mandate landed on paper in March 2026, and the city is still mid-process.

The timing matters. With 2028 Olympic infrastructure projects ramping up across Inglewood, Downtown and the Valley, every city agency tied to construction permitting is under pressure to maintain clean, fast, reliable digital records. Duplicate images in the system don't just waste server space — they've been linked to at least two documented errors in the Department of Building and Safety's Central District office on Spring Street, where field inspectors pulled outdated photos that showed pre-demolition conditions on lots already cleared for Olympic-adjacent development.

What LA Is Doing — And What It Isn't

The city contracted with a firm called Virteva in January 2026 to run a deduplication audit across the Bureau of Engineering's SharePoint and legacy FileMaker repositories. The contract, valued at $2.1 million over 18 months, targets roughly 1.4 terabytes of image data accumulated since digitization began in earnest around 2009. The audit uses perceptual hashing — a technique that compares images by visual fingerprint rather than file name — to catch duplicates that were renamed, re-exported or uploaded multiple times through different departments.

The Los Angeles County Department of Public Works runs a parallel but separate system, and the two entities are not yet coordinating on deduplication standards. That gap is significant. In East Hollywood and Boyle Heights, where the Bureau of Engineering and the county both photograph the same corridors for separate compliance purposes, overlap rates in some street segments reportedly exceed 60 percent.

Compare that to London, where Transport for London integrated a city-wide perceptual hash protocol into its asset management system back in 2023. By late 2025, TfL reported a 78 percent reduction in duplicate images across its 14,000-camera network archive. Tokyo's Bureau of Urban Development adopted a similar standard in 2022, building deduplication directly into upload workflows so images are checked at the point of ingestion rather than cleaned up retroactively. São Paulo, running a leaner system through its Secretaria Municipal de Urbanismo, mandated unique image IDs tied to GPS coordinates starting in 2021 — a low-tech but effective solution that cost roughly $180,000 to implement citywide.

The Retroactive Problem

Los Angeles, characteristically, is trying to fix the problem after the fact. That retroactive approach costs more and takes longer. The $2.1 million Virteva contract doesn't include any provision for reforming the upload workflows that created the duplicates in the first place — meaning the same problem will likely recur unless a second contract follows. The Bureau of Engineering did not respond to questions about whether such a follow-on contract is planned.

The LA City Archives on Pico Boulevard, which manages historical records separately from operational engineering files, completed its own smaller-scale deduplication project in October 2025, clearing about 22,000 redundant files from its photographic collection using open-source software. Archivist staff there have been pushing the Bureau of Engineering to adopt the same ingestion-point protocols that Tokyo and TfL use, according to a memo obtained through a public records request filed in April.

For residents and developers navigating the city's permitting system, the practical advice is straightforward: when referencing city-held images in permit applications — particularly in neighborhoods like Boyle Heights, Koreatown and Sylmar where infrastructure photography is dense — explicitly request the image file ID number, not just a description. That small step can prevent reviewers from pulling the wrong version of a photo that exists in three or four iterations in the system. The city's GeoHub portal, accessible at geohub.lacity.org, allows users to cross-reference permit photos against GPS-stamped records, which offers at least a partial workaround while the broader cleanup continues.