← The bench

02 · The label

Phenology in free text

Did the collector already write down what the model is trying to infer, and can you trust the tag?

Status
Running on the harvest now
Field
Thin — some prior work
Who wants it
Anyone training a phenology model, including us — and any institution that wants its phenology tags to be trustworthy.

Why it matters

Collectors routinely wrote in fruit, fl., sterile — an expert label generated in the field, sitting in pixels. But a keyword scan that fires on flower-bearing organs tags a sterile sheet as flowering. Running OCR on the parser's label region and mining only assertive phrases separates the real claim from the morphology note. On this harvest, most keyword tags checked this way turned out to be the descriptive trap, and OCR recovered real assertions the metadata never carried.

What it needs

The sheet parser (to find the label) and OCR (to read it). Both run now. Handwriting is the ceiling; printed and typed labels read cleanly.

State of the art

The keyword approach is common and quietly wrong. Reading the image and refusing descriptive text is not routinely done.

The working set

225 sheets in the harvest match this head’s starting slice

Open this slice in the catalog →

Next head

Type status detection