Fugleramme: Raspberry Pi e‑ink bird frame with real‑time local AI and 1800s illustrations

What Fugleramme Does and Why It Matters

Fugleramme turns ambient bird song into a living art gallery: a microphone feeds audio to the open‑source BirdNET‑Go classifier, which runs entirely on a Raspberry Pi, and each detected species is rendered as a historic 1800s illustration on an Inky Impression e‑ink display (or a web‑only kiosk). The system showcases how low‑cost edge AI can create tangible, aesthetically pleasing experiences without cloud services.

---

Core Architecture

Audio detection – A USB microphone streams to BirdNET‑Go, a locally‑run neural network that outputs species names.

Illustration pipeline – Fugleramme polls the BirdNET‑Go API, matches each species to a pre‑curated cut‑out from public‑domain plates, removes the background, and packs the images onto a virtual A4 page. Larger birds are placed toward the centre and scaled by real body mass (derived from AVONET data).

Display logic – The composed page is sent to an Inky Impression 13.3" e‑ink panel, which refreshes only when the set of detected birds changes, using a six‑colour dither to preserve the look of the original artwork. A parallel web kiosk serves the same view at :8080 and an admin UI at :8080/admin.

Optionality – The e‑ink panel is not required; the software can run headless in a Docker container and display the collage on any HDMI screen or remote device.

---

Artwork Collection

Fugleramme ships with 800+ cut‑outs covering 400+ species, all sourced from Scandinavian, British, and Central European natural‑history plates. The assets are fully public‑domain (CC BY‑SA 4.0 for the "classic" set) and manually curated—no AI‑generated art is used, although a few images have been lightly retouched with AI tools.

"Half the point of this project is showing off some amazing public‑domain natural‑history illustrations. Over 800 cut‑outs covering more than 400 species, every one taken from a real plate and hand‑curated for this project (no art is AI‑generated, though some has been retouched with AI)." – author comment

Coverage is strongest for the Nordics, British Isles, and Germany; extensions for broader Europe and North America are planned.

---

Getting Started

Raspberry Pi installation (single‑command)

curl -fsSL https://raw.githubusercontent.com/arnegiacomo/fugleramme/main/install.sh | bash

The script prompts for the BirdNET‑Go location, clones the repository, installs dependencies, and registers a systemd service. A reboot may be required on a fresh OS install.

Development mode (no hardware required)

uv sync                     # create a virtual environment
uv run fugleramme-fake-detector   # start a stub BirdNET‑Go on port 8090
uv run fugleramme-dev            # launch the service with hot‑reload on port 8080

Docker deployment

docker run -d -p 8080:8080 \
  -v fugleramme:/data \
  -e FUGLERAMME_DETECTOR_URL=http://birdnet.local:8080 \
  ghcr.io/arnegiacomo/fugleramme

The container persists data in /data and exposes the kiosk at http://<host>:8080.

---

Community Feedback and Ecosystem

• The project has attracted 2 189 up‑votes and 245 comments on Hacker News, indicating strong community interest.
• Commenters praise the blend of AI and heritage art, the low‑cost hardware, and the potential for personal gifts or commercial products.
• Several users note the high price of the 13‑inch e‑ink panel (≈ £229) as a barrier, while others suggest using alternative displays or repurposing existing e‑ink devices.
• A recurring theme is the desire for broader species coverage and for extensions such as individual bird identification or integration with other smart‑home displays.
• Some users point out prior art (e.g., Theodore R. AvianVisitors) and request attribution; the author acknowledges the inspiration in the comments.

---

Licensing and Attribution

• Code – MIT License.
• Bird detection – BirdNET‑Go model is CC BY‑NC‑SA 4.0 (non‑commercial) and credits Cornell Lab of Ornithology and Chemnitz University of Technology.
• Illustrations – Mostly CC BY‑SA 4.0; each style folder contains an ATTRIBUTION.md linking back to the original plates.
• Fonts – SIL OFL 1.1.
• Bird size data – AVONET body‑mass dataset, CC BY 4.0.
• Taxonomic aliases – OpenFauna mapping, CC BY‑SA 4.0.

---

Building or Buying a Frame

The author has assembled a few pre‑built frames and invites interested parties to contact him. The hardware bill of materials includes a Raspberry Pi 5, an Inky Impression 13.3" e‑ink panel, a microphone, and an A4‑sized frame; a full parts list and alternatives are documented in the repo’s docs/hardware.md.

---

Takeaways

• Fugleramme demonstrates that local AI inference can power delightful, low‑power IoT experiences without relying on cloud services.
• By coupling real‑time acoustic classification with public‑domain natural‑history art, the project creates a bridge between modern machine learning and historical scientific illustration.
• The open‑source nature, Docker support, and detailed documentation lower the barrier for hobbyists, educators, and makers to replicate or extend the system.
• Community interest suggests a market for aesthetically focused, AI‑enhanced home displays, potentially leading to commercial products that combine bird‑watching, art, and smart‑home integration.