Files
AvianVisitors/README.md
T
Twarner491 d3241f92cf Pre-share fixes from final review pass
Blockers
- http://birdnet.local/avian/ now serves the collage (added avian/index.html
  redirect to frontend/; previously the symlink target had no top-level
  index, so users saw a directory listing of api/ assets/ forwarding/
  frontend/ scripts/).
- Gemini model bumped from gemini-2.5-flash-image-preview (deprecated
  2026-01-15) to the stable gemini-2.5-flash-image. Pregen was non-
  functional out of the box.
- avian-mqtt.service: dropped %h (which expands to /root for system-mode
  units regardless of User=). Use WorkingDirectory=~ which DOES respect
  User=, plus a relative ExecStart.
- birdnet-api.php case 'lifelist' now aliases COUNT(*) AS n (was 'total'),
  matching what apt.js reads for atlas card counts. Atlas previously
  always displayed 0.

Bugs
- Retry-After parsing handles HTTP-date format without crashing.
- apt.js no longer fetches firstseen/timeseries (declared in state but
  never consumed). Two wasted round-trips per page load removed.
- update_birdnet.sh defaults branch=avian-visitors so the (opt-in)
  Sunday auto-update cron doesn't reset HEAD to upstream main and
  delete the avian/ overlay.
- recording.php: removed dead $BIRDS_JSON_CANDIDATES.

Polish
- Pregen --sleep default raised 4→6s (more headroom under free-tier RPM).
- en-dash in README BOM (~$35–80) replaced with hyphen for style match.
- Slop sweep across user-facing docs: no em-dashes, no en-dashes, no
  AI-slop word hits (robust/comprehensive/seamless/etc.).
2026-05-28 11:11:20 -07:00

3.4 KiB

AvianVisitors

A live bird collage from your window.

See it running at bird.onethreenine.net.

avianvisitors collage

BOM

Qty Description Price Link
1 Raspberry Pi (4B / 5 / Zero 2W) ~$35-80 Raspberry Pi
1 Micro SD Card (≥32 GB) ~$10 Amazon
1 USB lavalier microphone $14.99 Amazon
1 Pi power supply ~$10 -

Optional: a Gemini API key to restyle illustrations, an eBird API key to filter species by region.


1. Flash the SD card

Use Raspberry Pi Imager. Pick Raspberry Pi OS Lite (64-bit). In the customisation dialog set:

  • Username
  • WiFi SSID + password
  • Hostname: birdnet
  • Enable SSH with password auth

Plug the USB mic into the Pi. Place the capsule in a window or mount it outside. Boot.


2. Run the installer

ssh <your-username>@birdnet.local
curl -s https://raw.githubusercontent.com/Twarner491/AvianVisitors/avian-visitors/newinstaller.sh | bash

Clones this fork, installs BirdNET-Pi, symlinks the AvianVisitors overlay into the Caddy web root. Takes 20-40 minutes. Reboots when done.

Collage: http://birdnet.local/avian/. Stock BirdNET-Pi UI: http://birdnet.local/.


3. (Optional) Restyle the illustrations

The repo ships with 450 bundled illustrations. To restyle or add region-specific species:

export GEMINI_API_KEY='your-key'

# Re-render every species in BirdNET-Pi's model:
python3 ~/BirdNET-Pi/avian/scripts/pregen.py --labels ~/BirdNET-Pi/model/labels.txt --force

# Or filter to species observed in your eBird region:
export EBIRD_API_KEY='your-key'
python3 ~/BirdNET-Pi/avian/scripts/pregen.py \
  --labels ~/BirdNET-Pi/model/labels.txt \
  --ebird-region US-CA

Style lives in avian/scripts/prompt.template.md. Edit, re-run with --force.


4. (Optional) Forward off your LAN

See avian/forwarding/ for three independent recipes:

  • Cloudflare Tunnel for a public HTTPS URL.
  • Home Assistant REST sensor that exposes the latest detection.
  • MQTT bridge that publishes every new detection.

Repo layout

avian/                  # everything we add to BirdNET-Pi
├── frontend/           # static HTML/JS/CSS for the collage
├── assets/             # 450 bundled illustrations + cutouts + masks
├── api/                # PHP shims served by BirdNET-Pi's PHP-FPM
├── scripts/            # pregen.py + editable prompt template
└── forwarding/         # optional HA / MQTT / Cloudflare configs

Everything outside avian/ is upstream BirdNET-Pi.


License

CC-BY-NC-SA-4.0, inherited from BirdNET-Pi. Non-commercial use only. See the BirdNET-Pi README for full Cornell attribution.