INDUSTRY FIELD GUIDE
Stone fruit size and handling data for AI evaluation
A peach can look large in a close-up and still fall into a different size class when measured.
NAICS 111339 · Other Noncitrus Fruit Farming
A peach can look large in a close-up and still fall into a different size class when measured. A handoff decision may also need firmness evidence the image cannot supply. USDA publishes commodity-specific fruit grade standards. An orchard can use those distinctions to design an evaluation around two types of evidence. USDA fruit standards
Use an approved fruit sample. Pair an image with a calibrated scale marker and one without it. Label size under the training exercise’s method. Add an independent firmness observation where it is relevant and authorized. Give the model an out-of-order view and a separate firmness reading from an adjacent lot. Score its ability to make a measurement, cite the correct fruit, and request missing evidence. Choose one variety and handling stage for the initial sample.
Clear packhouse photographs and test readings with their owners. Keep private cultivar rights, commercial grading results, buyer names, and grower location data outside the example.
Describe an orchard inspection sample
Original Fiund exercise
A calibrated width and a separate firmness observation must belong to the same fruit
Original fictional peach annotations use practice length units. No images, real firmness tests, cultivar rights or handling recommendations are supplied.
| Record | Source | What it records |
|---|---|---|
| SFZ1 | Peach A calibrated view | Same-plane scale bar: 20 pixels=10 practice units. Reviewer-marked width spans 60 pixels. Fruit identity A confirmed. |
| SFZ2 | Other A view | Close-up without scale, captured earlier but sorted later in the file. Apparent width 90 pixels; no distance or magnification metadata. |
| SFZ3 | Firmness card | Reading F belongs to peach B from adjacent lot L2. No authorized firmness observation for A included. |
Question for the model
Which width can be derived for A, and do the close-up or B's firmness card establish A's size change or handling readiness?
Show the reference answer and checks
SFZ1's scale gives A a width of 30 practice units: 60 pixels times ten units per 20 pixels. SFZ2 lacks a scale and is earlier despite its file order, so its larger pixel span does not establish growth or a physical size difference. SFZ3 belongs to B and cannot supply A's firmness. Report the supported width and missing A-specific firmness evidence without deciding harvest or handling readiness.
- Comparing pixel widths across uncalibrated magnifications as physical growth.
- Assigning an adjacent lot's firmness reading to the photographed fruit.
Fields to preserve
- Fruit and lot identity
- Connects measurements to the specimen.
- Scale geometry and marked span
- Supports physical-unit conversion.
- Capture time and firmness-source link
- Separates chronology and measurement types.
Test another case
Give the close-up a scale of 30 pixels per ten units. Require its 90-pixel span also to resolve to 30 units despite the different image size.
Use owned approved fruit and original annotations. Clear packhouse images and tests separately; exclude buyer names, commercial grades and locations.
Discuss a contribution
Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.