OpenSurveillanceDB

An open, non-commercial civic database of visible public surveillance infrastructure.

About

OpenSurveillanceDB is an open, non-commercial civic database that documents visible public surveillance infrastructure. It maps where cameras and automated surveillance systems are installed in shared spaces, so the public can understand what observes them.

It is not a tool for watching, tracking or bypassing lawful surveillance: the project publishes data about cameras, not feeds — context, not profiles; transparency, not tracking.

Mission

Help people understand the systems around them.

The project maps visible camera infrastructure in shared spaces, so the public can see what is installed nearby. It documents public-facing equipment only. It is not a tool for watching, tracking or bypassing lawful surveillance.

What we document	What we do not collect	What the data cannot prove
Visible camera infrastructure, approximate location, type, source and a community status.	Camera feeds, credentials, private-home details, operational weaknesses, faces, licence plates or other personal data.	An absent record does not show that an area is free of surveillance. It only shows that no record is currently published there.

Principles

Five commitments shape every decision in this project, from the data model to the community actions.

• Free to use — no ads, no profiling, no paid features. The database and the software are public goods.
• Open source — the software can be inspected, reused and improved by the community.
• Open data with provenance — licensing and provenance are recorded for every published record, so journalism, research and civic use can rely on them.
• Privacy and safety by design — no private-home cameras, no sensitive operational details, no live-feed links. Faces and licence plates are removed before publication.
• Community accuracy first — reports are published immediately from verified accounts. The community keeps the directory accurate: confirming what is still there, flagging what is gone, and withdrawing what is problematic — under automatic, transparent thresholds.

What we deliberately do not do

The boundaries are as important as the mission. OpenSurveillanceDB will never become any of the following:

• No camera feeds — we document equipment, not footage. No live or recorded feed is ever linked or displayed.
• No tracking tools — the data cannot be used to watch, follow or profile people in public space.
• No evasion advice — the project does not help anyone avoid lawful surveillance and does not publish operational weaknesses.
• No private property — cameras that surveil private homes, and details of private life, are out of scope.

Features

• Interactive map (/mappa): OpenStreetMap tiles served through a same-origin proxy, viewport-first rendering, native grid aggregation at national zoom, camera field-of-view cones/circles above zoom 16, a keyboard-accessible marker popup with community actions.
• Public directory (/directory): text equivalent of the map — search, filters, A–Z index, CSV/GeoJSON exports, ?page= pagination.
• Record pages (/records/[id]): facts, location (address + coordinates), community status, a small interactive position map, a public event timeline, and an owner-only edit page.
• Community contributions (/segnala, /correggi): verified accounts submit cameras or corrections with map/coordinate location selection, reverse-geocode address prefill and a server-enforced duplicate gate. Community actions (useful / confirm / no longer there / problem / privacy) drive automatic, trust-weighted state transitions (ADR 0021).
• Accounts and auth: email+password with verification, passkeys (WebAuthn), and GitHub/Google OIDC sign-in — server-gated on configured providers; self-service erasure with de-attribution (GDPR art. 17).
• Public API: /api/cameras (JSON/GeoJSON/CSV), search, nearby, per-record actions and events, tile and geocode proxies — documented on the built-in /api-docs page, with a fail-open worker cache for public reads.
• Import pipeline: official open-data sources (city/regional/government portals, OpenStreetMap) via scripts/import/ — per-source adapters, a fail-closed licence gate, cross-source dedup, idempotency and per-batch attribution on the /fonti page.
• Bilingual interface: English and Italian, with structural parity enforced at compile time (ADR 0007).

Quick start

Requirements: Node.js 22.13 or newer.

npm install
npm run db:migrate   # apply the Drizzle schema migrations to a fresh local DB
npm run dev          # vinext dev server on http://localhost:3000

Open http://localhost:3000. The database starts empty — no rows are inserted at runtime. Two labelled illustrative pins for manual checks:

npm run db:seed

Useful commands:

npm run db:reset      # non-destructive local reset (moves state aside, re-migrates)
npm run db:generate   # regenerate a migration after changing db/schema.ts
npm run test          # build + full test suite
npm run lint
npm run build         # production build (dist/)
npm run db:provision  # provision real moderator/admin accounts

This is the beginning of the README. Read the rest at the original. Read the full README ↗