INDUSTRY FIELD GUIDE
Tortilla manufacturing data for AI stack and size checks
Tortilla production offers a practical visual task when a pack has the right count but the wrong size mix.
NAICS 311830 · Tortilla Manufacturing
Tortilla production offers a practical visual task when a pack has the right count but the wrong size mix. USDA's product-specification library provides a public procurement context. The proposed sample uses original dimensions and tolerances defined solely for the exercise. USDA product specifications
Stage circular paper props with a scale reference and two fictional order cards. Photograph one flat arrangement and one stack that hides several edges. Ask the assistant to identify which visible pieces match the requested size and whether the stack can be verified from the available view. Include a complete-count pack containing one smaller piece. A count-only answer should miss the intended discrepancy.
Annotate item boundaries, scale location, visible diameter and the evidence status of hidden items. The reviewer key should separate measurement from the order's declared requirement. Hold out a new arrangement and a rotated scale for evaluation. Score unsupported measurements independently from wrong counts, and include a case where an additional view is necessary.
A contributor tool could pair the order card with the inspection images at capture time. Use invented brands and simple props for a first pilot. Real product footage and customer requirements need permission before sharing. Establish whether the buyer needs geometric inspection, document matching or both before planning a larger dataset.
Discuss this original manufacturing sample with Fiund
Original Fiund exercise
A complete visible count can still include one undersized piece under the practice rule
Original circular paper props and an in-plane scale illustrate geometry. No tortillas, actual quality tolerance, customer order or food handling is involved.
| Record | Source | What it records |
|---|---|---|
| TSD1 | Mock order | Practice pack requires four circles. Each must be between 48 and 52 fictional length units inclusive under the original rubric. |
| TSD2 | Flat view | All four boundaries visible with calibrated in-plane scale. Reviewer measures diameters 49, 50, 51 and 44; exact item labels supplied. |
| TSD3 | Stack view | Same pack ID, top circle boundary readable. Three lower edges hidden; no independent full-stack measurements. |
Question for the model
What does the flat view establish about count and size, and could the stack view alone verify every diameter?
Show the reference answer and checks
TSD2 supports four visible pieces, satisfying count, while diameters 49, 50 and 51 meet TSD1 and 44 falls below its fictional lower bound. TSD3 cannot independently verify the hidden lower diameters from the top circle or a shared pack ID. Score count, per-item measurement and visibility separately rather than approve the mix from total alone. No real food tolerance or product-quality determination is implied.
- Passing a pack solely because its visible count is correct.
- Applying the top piece's measured diameter to hidden lower pieces.
Fields to preserve
- Pack and per-item identity
- Defines observations.
- Scale, diameter and practice bounds
- Supports geometry.
- Boundary visibility and unresolved items
- Limits verification.
Test another case
Rotate the scale and use a different arrangement while keeping it in the measurement plane. Require actual boundary measurements rather than a layout shortcut.
Use original paper props and arbitrary bounds. Exclude real product tolerances and customer materials without permission.
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.