Argus is an open-source, real-time interactive map of 229,000+ public traffic and CCTV cameras from government and commercial sources worldwide — highway DOT cameras, city traffic cams, and public webcams, aggregated from open data APIs and rendered on a GPU-accelerated map (with a 3D globe view built in). It's a React, TypeScript, and Deck.GL/MapLibre dashboard for exploring live camera feeds, HLS video streams, and static snapshot imagery by country, region, or city, backed by a Python scraping pipeline that keeps the dataset current.
It's two independent halves:
- Frontend (
src/) — a React + TypeScript dashboard. Reads a static JSON payload at runtime; it never talks to the scrapers directly. - Data pipeline (
scripts/) — a Python CLI that scrapes camera metadata from ~10 government/commercial sources into a local SQLite store, then exports it to the JSON payload the frontend reads.
- 2D tactical view — thousands of camera dots over a dark MapLibre basemap, rendered with Deck.GL for performance at scale. Points below a zoom threshold are pixel-binned in a Web Worker so panning/zooming stays smooth with all 229k nodes loaded.
- 3D globe view — the same data on a rotating globe (
react-map-gl+ MapLibre's native globe projection), toggleable from the HUD. - Live feed playback — HLS streams via
hls.jswhere available, falling back to a cache-busted static JPEG poll if the stream fails. A local dev-only proxy (scripts/server.py) resolves CORS andipcamlive://streams that a browser can't reach directly. - Filters, settings, and a country/sector browser for narrowing down the camera set, plus a "jump to random camera" action.
- In-app data sync — a Settings-panel button that (in development) triggers a scrape and re-export via the local control server, with live progress streamed into the UI.
Prerequisites: Node.js ^20.19.0 || >=22.12.0 (required by Vite 8), Python 3.9+.
Frontend
npm install
npm run dev # Vite dev server, http://localhost:5173Other frontend commands:
npm run build # tsc -b && vite build
npm run lint # eslint .
npm run preview # preview a production build locallyThere is no test suite configured (npm test does not exist).
Data pipeline (from scripts/)
cd scripts
pip install requests
python scraper.py --list # see all plugins
python scraper.py --all # run everythingCamera data lives in a SQLite store (scripts/data/cameras.db) and is exported to public/cameras.geojson plus the three-tier payload the frontend reads at runtime: cameras.core.json (map), cameras.labels.json (names, loaded behind it), and cameras.detail/ (one chunk fetched per camera opened).
Local dev control server (optional — powers the in-app "Data Sync" button)
python scripts/server.py # listens on http://localhost:8787, run alongside `npm run dev`Only the windy plugin needs a key — every other source is keyless. Create .env in the project root:
WINDY_API_KEY=your_key_here
VITE_WINDY_API_KEY=your_key_hereGet a free key at api.windy.com. Both vars must hold the same value — WINDY_API_KEY is used by Python, VITE_WINDY_API_KEY is exposed to the browser by Vite.
Argus/
├── src/
│ ├── App.tsx # Entire UI: HUD, settings, filters, feed panel, both map renderers
│ ├── binWorker.ts # Pixel-binning for the 2D map at low zoom (Web Worker)
│ ├── main.tsx # React entry point
│ └── assets/ # Demo screenshots
├── public/
│ ├── cameras.geojson # Full dataset, one feature per camera (also used by the pipeline)
│ ├── cameras.core.json # Positions/color/live flag — blocks first paint
│ ├── cameras.labels.json # Names — loads in behind core
│ └── cameras.detail/ # Per-camera detail chunks, fetched on open
├── scripts/
│ ├── scraper.py # CLI entry point — plugin registry, merge modes, maintenance passes
│ ├── store.py # SQLite schema + export to the three-tier payload
│ ├── server.py # Local-only control server behind the in-app "Data Sync" button
│ ├── scrapers/ # One plugin per source, organized by region:
│ │ ├── usa/california/caltrans.py
│ │ ├── usa/road511.py
│ │ ├── canada/bc/drivebc.py
│ │ ├── europe/uk/tfl_london.py
│ │ ├── asia/singapore/lta.py
│ │ ├── oceania/nz/nzta.py
│ │ ├── global/windy.py
│ │ ├── opencctv_bridge.py # Strategic bridge: syncs 200k+ nodes from OpenCCTV
│ │ ├── utils.py # build_feature() — normalizes every source to one schema
│ │ └── ...resolvers (ipcamlive, txdot, host_prober)
│ └── legacy/ # Retired scripts, kept for reference only
└── archive/ # Orphaned data files, kept instead of deleted
Computed from the live store (python scraper.py --stats), 229,308 cameras total.
By continent
| Continent | Cameras | Share |
|---|---|---|
| North America | 122,222 | 53.3% |
| Europe | 67,099 | 29.3% |
| Asia | 32,755 | 14.3% |
| Oceania | 4,772 | 2.1% |
| South America | 1,218 | 0.5% |
| Africa | 1,167 | 0.5% |
| Antarctica | 22 | <0.1% |
| Unclassified* | 39 | <0.1% |
By country (every country with 1,000+ cameras — 24 of them, covering 94.5% of the dataset)
| Country | Cameras | Share |
|---|---|---|
| United States | 110,912 | 48.4% |
| Japan | 15,127 | 6.6% |
| Canada | 10,884 | 4.7% |
| United Kingdom | 7,829 | 3.4% |
| Italy | 7,071 | 3.1% |
| Austria | 6,539 | 2.9% |
| Taiwan | 6,305 | 2.7% |
| Germany | 6,093 | 2.7% |
| Spain | 5,530 | 2.4% |
| France | 5,209 | 2.3% |
| Switzerland | 4,816 | 2.1% |
| Finland | 3,398 | 1.5% |
| Australia | 3,324 | 1.4% |
| Norway | 3,093 | 1.3% |
| Indonesia | 2,963 | 1.3% |
| Czechia | 2,849 | 1.2% |
| South Korea | 2,703 | 1.2% |
| Sweden | 2,684 | 1.2% |
| Vietnam | 2,107 | 0.9% |
| Poland | 1,893 | 0.8% |
| Slovenia | 1,548 | 0.7% |
| New Zealand | 1,370 | 0.6% |
| Russia | 1,365 | 0.6% |
| Hong Kong | 1,021 | 0.4% |
| Other (149 countries/territories) | 12,675 | 5.5% |
*A handful of rows (<0.1%) carry a malformed region tag from upstream sources rather than an ISO country code.
| Goal | Command |
|---|---|
| Camera counts by source | python scraper.py --stats |
| Run specific plugins | python scraper.py --plugins drivebc tfl_london nyc_dot |
| Run everything except Windy | python scraper.py --all --exclude windy |
| Target specific US states | python scraper.py --plugins road511_usa --states CO TN DE |
| Drop & refresh a source's cameras | python scraper.py --all --replace-source |
| Rebuild from scratch | python scraper.py --all --fresh |
| Run plugins concurrently | python scraper.py --all --parallel |
- Add Satellite Camereas from https://eumetview.eumetsat.int/static-images/latestImages.html
- Add Cameras from https://opencctv.org/api/hls-seg
MIT. This project is for educational and open-data visualization purposes only — all camera feeds are sourced from public, non-sensitive government or commercial APIs.