London's Duplicate Image Problem: The Hard Numbers Councils and Developers Don't Want You to See

More than one in five images uploaded to London's public planning portals contain duplicates, near-duplicates, or misattributed photographs, according to an audit of digital asset management practices reviewed by The Daily London. The figure, drawn from a Freedom of Information analysis of three London boroughs' planning databases, points to a systemic data hygiene problem that is quietly undermining housing policy at a moment when accuracy matters most.

The timing is not incidental. Keir Starmer's government has staked its domestic credibility on a planning reform agenda, and Mayor Sadiq Khan's housing teams rely on digitised records — site photographs, architectural renders, heritage images — to process applications and enforce conditions. When those records are polluted with duplicate files, the downstream effects range from the mundane to the genuinely costly.

What the Numbers Actually Show

The audit covered planning portals operated by Southwark Council, the London Borough of Tower Hamlets, and the Greater London Authority's own Development Database. Across roughly 340,000 uploaded image files examined over the period January 2023 to March 2026, an estimated 72,000 were flagged as exact or near-exact duplicates — images that appeared more than once under different file names, case reference numbers, or document categories. That translates to approximately 21 percent of the total image stock.

In Southwark, where regeneration activity around the Old Kent Road Opportunity Area has been intense, duplicate image rates ran higher than the average — closer to 28 percent in active case files opened after January 2024. The practical consequence is that planning officers must manually cross-reference documents to confirm whether an uploaded photograph depicts the site it claims to. At an officer time cost that council efficiency teams have internally valued at roughly £34 per hour, the aggregate burden across a single busy month is not trivial.

Tower Hamlets presented a different pattern. There, the duplication problem is concentrated not in officer uploads but in submissions from architectural and development firms. Files submitted through the borough's e-planning portal between April 2024 and December 2025 showed that several large commercial applicants — firms working on schemes around Whitechapel and Aldgate East — submitted identical photographic evidence under multiple planning reference numbers. In two documented cases, photographs of a site on Commercial Road were used to support applications for separate plots more than 400 metres apart.

Why Deduplication Has Proved So Difficult

The core technical issue is that London's planning infrastructure was not built with image-matching in mind. Most borough portals run on variants of Idox Uniform, a document management system widely deployed across English local authorities. Idox Uniform was designed to store and retrieve files, not to interrogate their visual content. Automated deduplication tools — the kind that cross-check pixel hashes or perceptual similarity scores — require a layer of software that most councils have not procured.

The GLA's Planning London Datahub, launched in 2019 to centralise data from all 33 London boroughs, ingests metadata rather than raw image files, which means the duplicate problem at borough level does not surface in the GLA's own dashboards. A borough can report 500 unique site photographs while holding 625 files, and the Datahub will record 500.

Digital transformation teams at several boroughs have flagged the issue to the Ministry of Housing, Communities and Local Government, which is currently consulting on a national digital planning standard. The consultation closes on 18 July 2026. Campaigners pushing for open, machine-readable planning data — including the charity Shelter and the Urban Land Institute's London chapter — have argued that image data quality deserves its own standard, separate from the text and tabular records that current proposals prioritise.

For residents and developers trying to understand what is actually happening on sites near them — on the Elephant and Castle estate boundary, say, or along the Rotherhithe riverfront — the duplication problem makes community scrutiny harder. If the photograph attached to a planning condition cannot be reliably tied to the site it is supposed to document, challenging a condition becomes that much more difficult.

Councils that want to get ahead of this before the MHCLG standard lands should start with a file-hash audit of their existing image libraries. It costs less than a single planning appeal to run, and the July 18 consultation deadline gives boroughs a narrow window to shape requirements before they are handed down from Whitehall.