Singapore Takes the Lead on Purging Duplicate Images From Public Records — But the Hard Work Is Just Far From Done

Singapore's Urban Redevelopment Authority quietly flagged the problem in its internal digital asset registers late last year: tens of thousands of duplicate photographs cluttering planning databases, some images stored four or five times across separate servers at Maxwell Road headquarters. The redundancy was costing storage budget, slowing retrieval, and — more critically — feeding inconsistent records into AI-assisted planning tools the agency had begun deploying across the island.

The issue is not unique to Singapore. But how this city is handling it, compared with peer cities in Asia and Europe, reveals both the ambition of its digital governance push and the practical friction that makes clean data far harder to achieve than any roadmap suggests.

Why Duplicate Images Have Become a Governance Problem

The proliferation of municipal digital archives accelerated sharply after 2020, when city governments worldwide pivoted to remote-work models and rushed unstructured data onto cloud platforms. For Singapore, the pivot aligned with the Smart Nation initiative, which has channelled billions of dollars into digitising everything from Housing Development Board flat records at Toa Payoh Hub to heritage documentation at the National Archives of Singapore on Canning Rise. The volume of photographic and visual data has grown faster than the frameworks to manage it.

Duplicate images are not merely a storage annoyance. When the same photograph appears twice in a planning system under different metadata tags — one correctly labelled as Buona Vista in 2019, another miscoded as Queenstown in 2021 — AI models trained on that data inherit the error. The Government Technology Agency, which oversees GovTech's data infrastructure from its Sandcrawler Building offices near one-north, has been working since at least early 2025 to introduce hash-based deduplication across select government datasets. Hash-based methods generate a unique digital fingerprint for each image file; if two fingerprints match, one copy is flagged for deletion or archival.

Tokyo's Bureau of Urban Development faced a comparable reckoning in 2023, when a government audit found redundant image files accounting for roughly 18 percent of total storage consumption across ward-level planning offices. The bureau's response was centralised and top-down: a single deduplication engine rolled out to all 23 special wards by March 2024. Amsterdam took a different path, contracting a Dutch civic-tech firm to build a federated system that lets each city district manage its own deduplication while feeding clean metadata to a central registry maintained by the Gemeente Amsterdam.

Singapore's Middle Path — and Its Gaps

Singapore appears to be threading between those two models. GovTech has pushed for centralised tooling through the Whole-of-Government Application Analytics platform, but individual statutory boards retain discretion over their own archives. That means the National Parks Board, which maintains extensive photographic records of green corridors from Bukit Timah Nature Reserve to the Rail Corridor, is running deduplication on a different timeline than, say, the Land Transport Authority.

The practical consequence is uneven progress. A 2025 report by the Infocomm Media Development Authority on Singapore's data quality benchmarks — released as part of the national Digital Connectivity Blueprint follow-up — noted that cross-agency image metadata consistency remained a work in progress, without specifying which agencies lagged. The IMDA did not release agency-level breakdowns publicly.

For residents and businesses interacting with government portals, the most visible sign of the problem is the occasional duplication of property images on the HDB Resale Portal, where the same flat photograph sometimes appears under slightly different listing identifiers. HDB has acknowledged the portal underwent backend data hygiene work in the first quarter of 2026, though the authority has not published detailed metrics on how many records were affected or corrected.

The comparison with Taipei is instructive. The city's Department of Information Technology completed a full deduplication audit of its municipal image library by December 2025, reducing stored visual data volume by 22 percent, according to figures published by the Taipei City Government. Singapore has not released an equivalent citywide figure.

For Singaporeans tracking the Smart Nation rollout, the practical takeaway is straightforward: submit feedback through the LifeSG app if government portals serve up duplicated or inconsistent property or planning images — it feeds directly into GovTech's data quality reporting pipeline. And watch the Digital Connectivity Blueprint's next progress update, expected in the fourth quarter of 2026, for whether cross-agency deduplication finally gets a measurable target attached to it.