São Paulo Confronts Its Duplicate-Image Problem — and the Rest of the World Is Watching

The Prefeitura de São Paulo confirmed this week that its urban mapping division had identified tens of thousands of duplicate images inside the city's official geographic information system — the same aerial and street-level photographs catalogued more than once, slowing searches, inflating storage costs and, in several documented cases, feeding incorrect data to the city's flood-monitoring tools along the Tietê River corridor. The disclosure, made in a technical bulletin circulated internally through the Secretaria Municipal de Urbanismo e Licenciamento, puts a concrete number to a problem that urban data managers have been flagging for at least two years.

The timing matters. São Paulo's drainage crisis has pushed city hall to lean more heavily on automated visual analysis — satellite snapshots, drone footage, street-level photography — to track illegal construction on floodplains and monitor the progress of the Programa de Drenagem Sustentável, the infrastructure initiative targeting the most flood-prone subprefeituras including Itaquera and Brasilândia. When the image library feeding those tools contains duplicates, analysts waste hours on manual verification rather than decision-making. A single major flooding event in the Zona Leste last March, which inundated stretches of Avenida Aricanduva, exposed exactly this bottleneck in real time.

How São Paulo's Approach Compares Abroad

Mexico City's Agencia Digital de Innovación Pública began a comparable deduplication project in late 2024, targeting the roughly 4 million street-level images inside its urban mobility database. The agency reported that perceptual hashing — a technique that generates a compact fingerprint for each image and flags near-identical pairs — cut redundant records by an estimated 34 percent within six months of deployment. Lagos State's geographic data authority attempted a similar clean-up in 2023 but stalled after procurement disputes delayed software licensing for more than a year. Bogotá's Instituto Distrital de Gestión de Riesgos y Cambio Climático integrated deduplication directly into its real-time camera feeds ahead of the 2025 rainy season, and city engineers there have cited faster incident response as a measurable result.

São Paulo's Municipal Data Governance Lab — housed since 2023 inside the Centro de Operações São Paulo on Rua Boa Vista, in the historic centre — is piloting a hybrid approach: automated perceptual hashing for aerial imagery combined with metadata cross-referencing for street-level shots captured under the city's partnership with mapping contractors. The lab is also coordinating with the Instituto de Pesquisas Tecnológicas, whose Cidade Universitária campus hosts the city's spatial data backup infrastructure. Early internal tests, according to the technical bulletin, reduced flagged duplicates in a test dataset covering the Pinheiros and Vila Mariana districts by approximately 28 percent — a figure the lab describes as preliminary pending a full audit scheduled for September 2026.

What the Fix Could Cost — and Who Pays

Storage alone is not a trivial line item. Cloud archiving for municipal geospatial data in Brazil has risen sharply since 2023, tracking broader infrastructure cost increases tied to the real's exchange-rate pressures. The city's current contract for geographic data storage, renewed in January 2026, runs to approximately R$4.2 million annually for the relevant tier of services. Eliminating a material share of duplicates could reduce that figure at the next renewal cycle — an argument that data managers have made directly to the Secretaria Municipal de Finanças as justification for the upfront software investment.

For residents and businesses, the practical downstream effect is less abstract than it sounds. Urban planning applications submitted through the SP Urbanismo portal on Avenida Paulista rely on cross-referenced map imagery to validate lot boundaries; duplicates in the underlying database have contributed to processing delays that some applicants have raised in public hearings at the Câmara Municipal. A cleaner image library means faster permit reviews — a chronic sore point for developers and small property owners alike.

The September audit will determine whether the pilot scales city-wide. If it does, São Paulo's data governance team intends to publish the methodology as open-source documentation, making it available to other Brazilian municipalities. Whether cities like Fortaleza and Manaus, which run their own geographic information systems at significantly smaller scale, can adapt the tooling to their own infrastructure is a question the lab says it expects to answer by the end of 2026.