seeni

Attributions

Seeni is built on photos, data and software made by others. This file credits them and states the licenses that apply. See README.md for the license of Seeni’s own code, model and species data.

Sections marked pending are filled in by the phase in brackets. Nothing in them is in use yet.

Training images

Pending (#4, #11).

The model is trained on photos shared through GBIF under CC0 or CC-BY licenses: mushrooms, and for the unknown class muu also other fungi and non-fungi such as moss, cones, leaves and plants (#7). Every photographer is credited, whatever the class. Each photo’s license is taken from GBIF multimedia.txt, not from the occurrence record. The photos are not distributed in this repository or with the app, but the model is derived from them.

Images committed to this repository

Any image in this repository that is not Seeni’s own work must be CC0 or CC-BY and is listed here. This covers test fixtures (any tests/fixtures/ folder) and images under web/, species/ and docs/, such as verification check pictures.

File Author License Source
none yet      

Species data sources

Pending (#3). data-meta/lajilista.csv and species/species.json are compiled from the sources below. Each source’s license and required credit is recorded here before release.

Source Used for License
GBIF Backbone Taxonomy taxonKeys To be checked
Laji.fi (FinBIF) taxonomy Finnish names To be checked
Ruokavirasto list of commercial mushrooms Species selection To be checked

Pretrained weights

Pending (#5, #6). Check each license from the model card before release.

Model Used for Source License
DINOv2 Baseline only, not shipped (#5) facebookresearch/dinov2 To be checked
EfficientNet-B0 Starting point of the shipped model (#6) timm To be checked

Software

Libraries shipped with the web app:

Library License
Pending (#10): onnxruntime-web MIT
Pending (#10): vite-plugin-pwa and Workbox MIT

Tools used to build the app and train the model, not shipped:

Tool License
PyTorch BSD-3-Clause
torchvision BSD-3-Clause
timm Apache-2.0
Vite MIT
TypeScript Apache-2.0