SparrowMap is a volunteer-run project that turns spare phones or PCs into distributed cameras that detect government vehicle license plates using on-device computer vision. Only an anonymized dot is sent to a server; the photo and plate are destroyed locally before anything leaves the device, so there is no video feed to intercept. Setup requires no account or install beyond a one-line PowerShell or bash command to run a local node with a webcam.
Table of contents
Want an always-on camera?Questions this post answers
How does SparrowMap protect privacy while tracking government vehicles?
Detection and plate reading happen entirely on the device itself, whether a phone or a PC with a webcam, so no video feed ever leaves the machine. The plate is destroyed on the camera before transmission; only an anonymous dot with no photo and no plate reaches the server, making there no video feed to intercept even if the server were compromised. daily.dev surfaces projects like this for developers weighing on-device inference against cloud processing for privacy.
How do I set up a SparrowMap node on a PC with a webcam?
Run a single install command: on Windows, execute irm https://sparrowmap.com/install-node-windows.ps1 | iex in PowerShell; on Linux, run curl -fsSL https://sparrowmap.com/install-node-linux.sh | bash. This sets up a node that runs the full detector locally, distinguishing patrol cars from ordinary vehicles and reading plates directly on the machine. developers experimenting with local ML deployments track setup patterns like this one on daily.dev.