Dhaka's Digital Archives Are Drowning in Duplicate Images — The Numbers Tell a Damaging Story

At least 40 percent of all image files stored across Dhaka City Corporation's digitisation projects are believed to be redundant duplicates, according to internal assessments reviewed by technicians working on the municipality's data consolidation drive in early 2026. The figure points to a systemic failure in how Bangladesh's capital handles its rapidly expanding digital photo and document archives — a problem with real costs attached to every wasted gigabyte.

The issue has come into sharp focus this year because Dhaka South City Corporation (DSCC) and Dhaka North City Corporation (DNCC) are both mid-way through separate smart-city infrastructure programmes, each generating thousands of scanned records, surveillance stills and civic document images every week. Without a unified deduplication protocol, those images pile up in parallel storage systems, often identically filed under different naming conventions. The result is ballooning cloud expenditure and retrieval delays that affect everything from land registry queries in Motijheel to birth certificate issuance in Mirpur.

What the Data Actually Shows

Storage costs are not abstract. Cloud hosting for government data in Bangladesh runs at roughly Tk 3,500 to Tk 6,000 per terabyte per month depending on the vendor tier, based on pricing ranges circulated by local ICT procurement offices. When duplicate image files inflate a dataset by 30 to 50 percent — a range consistent with global benchmarks for unmanaged institutional archives — the direct financial waste on a 200-terabyte government archive can exceed Tk 4 lakh per month. Multiply that across a dozen agencies and the annual figure enters the crores.

Bangladesh's e-commerce sector in Dhaka compounds the problem. Platforms operating out of Panthapath and Tejgaon industrial corridors — where several logistics-tech and online retail firms cluster their back-end operations — routinely duplicate product images across multiple category pages, microsite variations and promotional campaigns. An image audit conducted by a Bangladeshi software consultancy in late 2025 found that one mid-sized Dhaka-based fashion e-tailer was storing the same product photograph in an average of eleven distinct copies across its content delivery network, each slightly resized or recompressed. That single platform's image directory had grown to 18 terabytes, of which a deduplication pass trimmed roughly 7 terabytes.

The ICT Division under Bangladesh's Ministry of Posts, Telecommunications and Information Technology has flagged digital asset management as a priority under the Smart Bangladesh 2041 framework, but no mandatory deduplication standard has yet been gazetted for public-sector institutions. The Bangladesh Computer Council, headquartered in Agargaon, runs training programmes for government IT staff, but image lifecycle management has not featured prominently in the published 2025–26 curriculum.

Local Organisations Starting to Act

Some institutions are moving without waiting for a top-down mandate. The Dhaka-based digital archiving project run through the National Archives of Bangladesh, located off Sher-e-Bangla Nagar, piloted a hash-based deduplication tool on its historical photograph collection in the first quarter of 2026. The pilot reportedly cleared 22 percent of redundant files from a test batch of 500,000 images within 72 hours of processing — a result that impressed enough that a broader rollout is under consideration.

Private media organisations are watching closely. Several Bangla-language online news portals headquartered near Karwan Bazar have begun auditing their own photo CMS databases after storage invoices spiked in 2025, partly because photojournalists' submissions were being ingested multiple times through different upload channels without any automatic duplicate check at the intake stage.

For organisations still operating reactive, ad-hoc storage policies, the practical path forward involves three concrete steps: adopting perceptual hashing tools that identify near-duplicate images even when file names differ, setting a single canonical storage location before any new digitisation project begins, and running quarterly audits rather than waiting for a budget crisis to force a cleanup. The Bangladesh Association of Software and Information Services, based in Karwan Bazar, has indicated it may publish guidance on the issue before the end of 2026, though no formal publication date has been confirmed. The cost of doing nothing, the numbers already on the table suggest, is no longer negligible.