Kōkiri Lab
Kōkiri Lab
MaramaTrap

About the science

Why moths?

Moths are sensitive to land use, vegetation, artificial light, and predator pressure. Recording what flies at your place — with honest habitat notes — turns a photo into biodiversity evidence, not just a gallery (spec 01).

Community submissions use the same 13-variable habitat protocol as the UCOL research strand so school investigations and open uploads stay comparable.

Model limits (stated plainly)

  • Order-level identification today — family-level training is in progress (spec 03 §6).
  • Confidence tiers are always shown in words, not colour alone.
  • “Not sure” is a valid outcome — better than a wrong guess.
  • Your family corrections on this device help improve future models.

17 moth families we are learning

The classifier is moving toward family-level IDs. When you correct a suggestion, pick from this list (spec 05).

  • Family-level only for now — not species claims (spec 03 §7). Corrections on your contributions help future training.

Also try community science

The same fair-testing habits work across platforms — honest notes, repeat visits, no street addresses.

All four platforms →

Data & privacy

  • Photos: EXIF/GPS stripped from images before storage (existing privacy fix).
  • Location: Optional browser GPS stored privately; public map uses ~1 km rounding (spec 01 §1).
  • Identity: Browser-local contributor ID — no account required (spec 04).
  • School vs community: MaramaTrap uploads are separate from consent-gated Kōkiri Lab investigations.
  • Research use: Non-commercial community science; aggregate data may support model training and open research under CC-BY-NC principles elsewhere in this project (spec 03 §6).
  • Open data: Anonymised aggregate export is planned — CSV/JSON for researchers and students (not yet live).