Local-AI e-ink frame IDs garden birds by ear, draws them as 1800s plates

A Norwegian developer has released Fugleramme, an open-source project that turns a Raspberry Pi 5 and a 13.3-inch Inky Impression e-ink panel into a living natural-history poster. A microphone feeds BirdNET-Go, a fully local classifier built on Cornell Lab’s BirdNET model, which identifies species by their calls. Fugleramme polls that classifier’s API, matches each detected bird to a hand-cut public-domain illustration, and composes them onto a textured paper page — larger birds sized by body mass toward the center, an empty perch when nothing is heard. The display only redraws when the set of birds changes, which suits the slow refresh of e-ink.

The project leans hard on curation rather than generation: over 800 cut-outs spanning more than 400 species, each lifted from real Scandinavian, British, and central European plates. Notably, none of the artwork is AI-generated, though some has been retouched with AI — a deliberate contrast to several similar bird-frame projects that use generated imagery. Coverage is strongest across the Nordics, British Isles, and Germany, with broader European and North American plates planned. All detection runs on-device with no cloud dependency.

Installation is offered several ways — a one-line install script that sets up a systemd service, a Docker image, and a compose file that stands up BirdNET-Go alongside it. The e-ink panel is optional; without it, Fugleramme serves the same view as a web kiosk over HDMI or to any device on the network. Code is MIT-licensed, while the bundled detection model and artwork carry their own non-commercial and share-alike terms. The author notes it’s still early-stage and welcomes contributions, especially new artwork and species coverage.