Wednesday, June 10, 2026 · 20 species observed
Selected from 7 frames across 7 visits.
Click any thumbnail to make it the Moment.
Wednesday, June 10. Past week's average: 29/day. Day still in progress.
Wednesdays typically see 34 visits.
Today vs the past 7 days. Knowledge layered on top.
What happened in each window.
Mostly American Crow (6 of 8). House Finch also dropped by.
The cast of characters, with trends and recent activity.
When each species peaks.
No two species really share the fountain. The closest pair — Chestnut-backed Chickadee and House Sparrow at 34% — still spend most visits apart. Overlap ranges from 0% to 34%, with most pairs in the 0–14% band. 31 pairs stay apart entirely — most notably American Goldfinch and American Robin.
66 species pairs analyzed in total.
Statistically notable observations and likely next visitors.
About these identifications
Species: Made by a local AI model and may contain errors. Identifications are reviewed and corrected over time.
Behaviors: Some are reliable (bathing, drinking are clear from video). Others like preening or vocalizing are often guesses — the model fills in what a species typically does at a birdbath rather than what it actually observed.
Sex and age: Easy calls like adult male House Finch are reliable. Female vs juvenile distinctions and most other sex/age calls are rough approximations.
How Birdwatch works
Hardware: A small camera detects motion and records short clips. An AMD GPU handles inference locally via ROCm.
Infrastructure: Everything runs in Podman containers on a self-hosted Linux server. Ollama runs the qwen2.5vl:7b vision language model locally. No cloud inference.
Pipeline: camera clip lands in an ingress folder. A systemd path watcher detects it and triggers processing. The coordinator extracts 5 frames and sends them to Ollama for identification. Results are written to JSON, the clip moves to archive, and a reporter script reads all JSONs, applies corrections, and generates the HTML report. A scheduled timer pushes the report to GitHub Pages.
Corrections and accuracy over time: Clips are reviewed via a LAN-only correction UI. When a bird ID is corrected, it gets saved and triggers automatic description regeneration and report rebuild. Before analyzing any new clip, the model receives a list of every species previously verified at this fountain. This should reduce repeated mistakes on birds the model has already gotten wrong once — but it is a nudge, not a guarantee.
