San Francisco Is Quietly Building a Blueprint for Fixing Its Duplicate Property Image Problem — Here's How It Stacks Up Against London and Seoul

San Francisco's Planning Department is running duplicate property images through an automated scrubbing pipeline, a quiet but consequential effort to clean up the tangled visual records that underpin everything from building permit approvals to affordable housing listings. The city confirmed the program is active across its public-facing property database, which covers more than 200,000 parcels citywide.

The stakes are higher than they might look. When the same photograph appears multiple times in a city's housing inventory — whether through sloppy data entry, vendor duplication, or algorithmic errors — it can distort unit counts, slow permit processing, and mislead prospective tenants searching for homes in neighborhoods like the Tenderloin or the Outer Sunset. In a housing market where the median asking rent for a one-bedroom hovers around $2,900 a month, any friction in the listing pipeline has real consequences for real people.

The timing matters because San Francisco is simultaneously pushing through a housing production emergency declared in late 2024 and trying to modernize the Department of Building Inspection's aging digital infrastructure. Duplicate imagery isn't a glamorous problem, but it sits at the intersection of both challenges.

What San Francisco Is Actually Doing

The Office of Housing and Community Development, which administers programs including the Small Sites Program that has preserved dozens of rent-controlled buildings in the Mission District and Chinatown, began flagging duplicate image entries in its internal property documentation in early 2025. Staff there began working with the city's Department of Technology to apply hash-matching software — a standard tool in digital asset management — to identify identical or near-identical image files across the database.

The San Francisco Planning Department's DataSF portal, which publicly hosts parcel data updated on a rolling basis, is also involved. City technology staff have been cross-referencing imagery attached to conditional use permits and environmental review filings, where duplicate photographs have occasionally caused processing delays by inflating document package sizes beyond the system's upload limits.

Meanwhile, Tenderloin Housing Clinic, a nonprofit that manages supportive housing for formerly homeless residents and operates properties on Eddy Street and Turk Street, has dealt with the downstream version of this problem: images from one building appearing in intake records for another, creating confusion during placements. The organization has pushed for cleaner data feeds from city systems.

How San Francisco Compares to London and Seoul

London's approach is instructive. The Greater London Authority, which manages the London Development Database tracking housing approvals across 33 boroughs, began a formal duplicate-image audit program in 2023. By mid-2024, the GLA had eliminated more than 14,000 duplicate image records from planning application files, according to reporting from local government trade publications. The city contracted with a specialist data quality firm and gave the process a two-year runway with dedicated budget allocation.

Seoul went further. The Seoul Metropolitan Government integrated AI-powered image deduplication directly into its public housing application portal in 2023 as part of a broader smart-city initiative. Applications for units in city-managed complexes in districts like Nowon and Mapo are now automatically screened for duplicate supporting photographs at the point of submission, cutting processing times measurably.

San Francisco's effort, by contrast, is more improvised. There is no dedicated budget line visible in the fiscal year 2025-26 city budget for image deduplication specifically, and the work is being absorbed by existing staff rather than contracted out. That makes it cheaper but slower, and more dependent on staff bandwidth that tends to compress when higher-profile crises — fentanyl response, encampment clearances — absorb department energy.

The practical difference shows up in timelines. London's two-year program had a defined endpoint and measurable output. San Francisco's effort has neither, at least not publicly.

For residents and housing advocates, the clearest near-term signal will come when DataSF updates its parcel imagery records later this year — city technology staff have indicated an updated data refresh is planned for the third quarter of 2026. If duplicate image counts drop significantly in that release, it will suggest the pipeline is working. If not, advocates say they will push for the kind of formal, budgeted program that London built. Watching the Q3 DataSF update closely is the best way for anyone tracking housing transparency to gauge whether the city's informal approach is holding up.