Hong Kong's Digital Archive Push Forces Hard Choices on Duplicate Image Replacement

Hong Kong's public institutions are sitting on a problem they can no longer defer. Across government databases, hospital records systems, and the city's major cultural repositories, duplicate digital images — scanned documents, archival photographs, patient intake files — have accumulated over decades of piecemeal digitisation, creating storage bloat, retrieval errors, and, in some cases, version conflicts that affect operational decisions. The question now is not whether to replace them, but how, at what cost, and under whose authority.

The urgency sharpened after the Innovation, Technology and Industry Bureau flagged digital asset deduplication as a priority area under the city's ongoing smart city blueprint review, which is expected to produce updated policy guidelines by the fourth quarter of 2026. For institutions that have delayed their own internal audits, the window to align with government standards before those guidelines lock in is closing fast.

What the Backlog Looks Like on the Ground

The scale is uneven across sectors. The Hong Kong Public Libraries network, which operates 72 branch locations citywide including the flagship facility on Pitt Street in Yau Ma Tei, has been migrating its visual catalogue to a centralised cloud platform since 2023. Staff at several branches identified significant duplication rates in scanned heritage photograph collections — particularly images originally processed across different borough-level systems that were merged without deduplication protocols. The Hong Kong Film Archive in Sai Wan Ho faces a related but more technically complex version of the same issue, where multiple master-quality scans of the same film frames exist in different compression formats, complicating preservation decisions.

Hospital Authority systems present a different set of stakes. Patient imaging files — X-rays, MRIs — processed at facilities including Queen Mary Hospital in Pok Fu Lam and Prince of Wales Hospital in Sha Tin have historically been stored in siloed departmental servers. Where duplicates exist, they raise not just storage costs but clinical risk, particularly when different versions of the same scan carry different annotation histories. The Authority has not publicly disclosed the scope of its deduplication exercise, but procurement notices published on the Government Electronic Trading Services portal this year indicate active tendering for medical image management software upgrades.

Decisions That Cannot Wait Much Longer

Three choices will define what the replacement process actually looks like. First, institutions must decide whether to pursue automated AI-assisted deduplication or manual curatorial review. Automated tools are faster and cheaper upfront — enterprise-grade solutions from vendors operating in the Hong Kong market have been quoted at between HK$200,000 and HK$800,000 for mid-sized deployments, depending on dataset volume — but they carry a documented risk of incorrectly flagging historically distinct images as duplicates, particularly in archival collections where near-identical photographs carry different metadata provenance.

Second, there is the question of retention policy. Deleting a duplicate is irreversible. Institutions under the purview of the Government Records Service, which administers the Public Records Ordinance, must ensure that any image retired from a primary database has a verified master retained in compliance with retention schedules. The Records Service has published guidance on digital records management, but sector-specific application — particularly for visual media in arts organisations like the Hong Kong Museum of History in Tsim Sha Tsui — remains inconsistently interpreted.

Third, and most consequentially for budget planning, is the question of whether replacement images must be sourced, re-licensed, or reconstructed from original materials. For institutions holding collections under expiring licensing agreements with overseas photographic agencies, the duplicate replacement exercise has surfaced a secondary problem: some originals were never fully rights-cleared to begin with.

Institutions that act before the Innovation, Technology and Industry Bureau guidelines are finalised in late 2026 risk building systems that require immediate retrofitting. Those that wait risk being frozen out of the first tranche of government co-funding for digital infrastructure, which is expected to be structured around compliance with whatever standards emerge. The practical advice from technology procurement specialists working with public-sector clients in Central and Wan Chai is consistent: complete your internal audit now, document every legacy duplication decision, and do not commit to a vendor contract that does not include a compliance update clause tied to the new guidelines.