Museo Avellonia
The client
Museo Avellonia is the online home of Göran Sjöberg's private collection of Lepidoptera (butterflies and moths), kept at Villa Avellonia: a structured archive of its specimens, built for researchers and the public. Prof. Markus Franzén commissioned the site, which sits alongside two sister sites built for him, the Quercus Project and Biodiversa.
Its audience is expert researchers and interested members of the public who want to look up a specimen and its taxonomy. The site has to work two ways at once: a public reference that reads clearly to a visitor with no training, and a curator's workbench for turning a photograph of a specimen box into structured, searchable records.
Digitising a specimen box
A specimen box can hold dozens of individually pinned specimens, each with its own printed label. The curator dashboard imports a photograph of each box from a shared Google Drive folder and runs it through a computer-vision model, the Segment Anything Model (SAM), which proposes a region for each specimen it finds. The curator confirms, adjusts or redraws a region, and adds any the model missed, rather than drawing every one from scratch.
Those regions are stored in three nested, percentage-based coordinate spaces rather than pixel coordinates: a specimen's position within the box, a label's position within the specimen, and a crop's position within the label. Because none of the three is tied to a fixed pixel grid, the same region still lines up correctly when the underlying image is served at a different size, from a full deep-zoom view down to a thumbnail.
The video (42 seconds, no sound) shows one specimen being digitised: marking its label on the photo of the box, checking the crop, approving the image with its background removed, completing the record and saving, then moving on to the next specimen.
At a glance
- Client
- Prof. Markus Franzén, for Göran Sjöberg's collection at Villa Avellonia
- Collection
- Lepidoptera (butterfly and moth) specimens, their taxonomy, localities and associated media
- Platform
- Drupal 11
- Site
- museo.avellonia.se
- Status
- Built; the move to the collection's own domain is still to come
- MADDev products
- Species External Entity, External Entities
What we built
- A structured content model for specimens, their taxonomy, collecting localities and associated media
- The same species enrichment as the Quercus Project and Biodiversa sites, from Catalogue of Life, GBIF, iNaturalist, the IUCN Red List and Wikipedia
- Deep-zoom viewing of specimen-box images, so a visitor can move from a whole box down to the detail of one specimen without loading one oversized image
- A side panel that shows a specimen's data next to the image it belongs to
- A Darwin Core Archive export, the standard exchange format for biodiversity data, which also guards against duplicate species records
Host-plant records
Lepidoptera taxonomy terms on the site carry host-plant records: what a caterpillar of that species is known to feed on. The most direct route would have reused a broad interaction database already integrated into the site, covering relationships across many taxa. We chose not to: that database's aggregation silently drops the per-record weighting used to rank a host plant's importance, along with two columns, on damage and on how and where a specimen was reared, that only the Natural History Museum's own HOSTS dataset records.
So the site queries HOSTS directly, and falls back automatically to the broader database for taxa, mostly tropical and Indo-Pacific species, that HOSTS does not cover. Results appear through the same related-species display used elsewhere on the site.