E-ink Bird Frame Uses Local AI to Draw 1800s-Style Illustrations
Raspberry Pi-powered device identifies birds by sound and displays them as vintage illustrations
1Real-time bird identification
The system uses BirdNET-Go's audio classifier to detect bird species through a microphone. When birds are heard, it matches them to a database of over 800 hand-curated 1800s illustrations. The Raspberry Pi-powered device updates an e-ink display only when new birds are detected, creating an energy-efficient nature display. Currently covers 400+ species primarily found in Scandinavia, Britain and central Europe.
2Vintage illustration collection
All artwork comes from public-domain natural history plates, manually cut out and retouched for the project. The system arranges birds on a textured paper background with larger species positioned centrally based on body mass data from AVONET. The collection focuses on European species but plans to expand coverage for North American birds. No AI-generated art is used in the project.
3Flexible hardware setup
While designed for Raspberry Pi 5 with an Inky Impression e-ink panel, the system can run without specialized hardware. Users can deploy it as a web kiosk on any display or access it from network devices. The project provides detailed hardware recommendations including microphone options and alternative components for different setups and budgets.
4Local processing advantage
All audio processing and species identification happens locally on the device without cloud dependency. The system polls BirdNET-Go's API which can run on the same machine or elsewhere on the network. This approach maintains privacy while allowing flexible deployment options. The admin interface enables configuration of display preferences and update settings.
5DIY-friendly deployment
Installation can be done via a single curl command that sets up dependencies and configures the systemd service. Docker containers are available for simpler deployment on Linux systems with USB microphones. The project maintains comprehensive documentation including full install guides for blank SD cards and hardware recommendations.
6Artistic inspiration
The visual design was inspired by a WWF poster by Axel Thorenfeldt and the concept of live bird displays from AvianVisitors. The creator emphasizes that while the detection uses AI, the artwork remains authentic historical illustrations. Each style folder includes attribution linking back to original plates and sources, with most artwork under CC BY-SA 4.0 license.
7Community contributions
The open-source project actively encourages contributions, particularly for artwork expansion, documentation fixes and detection improvements. The code is MIT licensed while detection components use CC BY-NC-SA 4.0. The system incorporates multiple open datasets including eBird taxonomy and OpenFauna's taxonomic aliases.
8Future development
Currently in early development, the project plans to expand bird coverage for broader European and North American species. The creator offers prebuilt units for those not wanting to DIY while maintaining the open-source approach. Future updates may include more display customization options and improved detection algorithms.