Singapore's Duplicate Image Problem: The Key Decisions Ahead for Government Records and Digital Archives

Singapore's government agencies and statutory boards are sitting on a growing problem: digital image archives riddled with duplicates, some repositories carrying redundancy rates that IT auditors have flagged as pushing unnecessary storage costs into the millions of dollars annually. The question now is not whether to act, but how — and which vendor, which standard, and which timeline will define the answer.

The urgency sharpened this year after the Smart Nation and Digital Government Office rolled out its updated data governance framework, which explicitly requires agencies to achieve measurable reductions in redundant digital assets by the first quarter of 2027. That deadline gives procurement teams at places like the Housing and Development Board and the National Library Board roughly nine months to make binding decisions on deduplication tools and workflow overhauls. Miss the window, and agencies risk non-compliance reviews and potential budget penalties in the next spending cycle.

Why Duplicate Images Accumulate — and What It Costs

The mechanics are mundane but expensive. When a planner at HDB's Toa Payoh Hub uploads a site photograph, the same image often gets re-uploaded by a contractor, archived again after revision, and stored once more when a project file is closed. Multiply that across hundreds of active projects and thousands of staff across campuses from Queenstown to Tampines, and the redundancy compounds fast. Industry benchmarks suggest that in large public-sector image libraries, duplicate and near-duplicate files can account for between 30 and 60 percent of total stored data, though the precise figure for any Singapore agency has not been disclosed publicly.

Storage is not free. Government-grade cloud storage procured through GovTech's centralised contracts carries real per-gigabyte costs, and agencies that overshoot their allocated tiers face supplementary charges. Beyond money, bloated archives slow search retrieval — a practical problem for archivists at the National Archives of Singapore on Canning Rise, where staff routinely pull historical images for public and policy use.

The challenge is compounded by image variety. Satellite imagery ingested by the Singapore Land Authority, body-camera footage managed by the Singapore Police Force, and construction-site photographs logged by the Building and Construction Authority all sit in separate silos with different metadata standards. A duplicate in one silo is invisible to another, making system-wide deduplication harder than it sounds.

The Decision Points That Will Define the Outcome

Three choices will carry the most weight over the coming months. First, agencies must decide between perceptual hashing tools — software that identifies near-identical images even when file names or formats differ — and more expensive AI-powered similarity detection that can flag contextually related images for human review. GovTech has been piloting both approaches, but no agency-wide procurement decision has been announced as of July 2026.

Second, there is the question of centralisation. A unified deduplication layer shared across agencies would reduce costs and standardise metadata, but it requires inter-agency data-sharing agreements that have historically taken time to negotiate. A federated model, where each agency runs its own deduplication engine, is faster to deploy but produces inconsistent results and misses cross-agency duplicates entirely.

Third, and most sensitive politically, is the question of deletion authority. Removing a duplicate from a government archive requires someone to sign off that the retained copy is authoritative. Lawyers at the Attorney-General's Chambers have previously advised that certain categories of official imagery carry evidentiary status and cannot be deleted without formal sign-off under the Evidence Act. Clarifying that boundary — what can be automatically purged versus what needs manual review — is a legal workstream that agencies cannot shortcut.

For citizens and businesses, the practical stakes are real. Developers submitting building plans through the GoBusiness portal attach image documentation that feeds into these archives. Faster retrieval and cleaner records mean faster approvals. The Urban Redevelopment Authority has flagged digital archive quality as one variable affecting its target to process straightforward development applications within the timeframes it has publicly committed to. Getting deduplication right is, in that sense, not a back-office IT problem. It is a service-delivery issue with a ticking clock attached.