INDUSTRY FIELD GUIDE
Miscellaneous textile product data for AI assembly evidence
This broad industry benefits from a concrete example.
NAICS 314999 · All Other Miscellaneous Textile Product Mills
This broad industry benefits from a concrete example. An original textile-filter assembly exercise can test layer order and orientation. INDA's nonwovens material provides context for the underlying material category, while all assembly rules in the proposed pilot are self-authored. INDA nonwoven context
Create a safe tabletop assembly from colored fabric layers and a paper frame. Define a fictional instruction with a marked upstream side. Photograph the stack before closure and the finished assembly afterward. Ask the model which required features remain visible and which need the earlier view. Include a correctly layered stack inserted backward into the frame.
The key should preserve layer identity, orientation and the evidence available at each stage. Evaluation can change colors and frame shape while retaining explicit instruction labels. Score orientation errors separately from incomplete visibility. Do not infer filtration performance from a correct-looking assembly, because the exercise contains no performance measurement.
Use original instructions, common materials and neutral identifiers. Actual customer designs, technical fabrics and test reports may have restrictions beyond ownership of a photograph. A licensing proposal should name this specific assembly task and its reviewer key. Confirm the buyer's requested task before treating the entire miscellaneous-textile category as a collection opportunity.
Discuss this original manufacturing sample with Fiund
Original Fiund exercise
Spot an orientation error without guessing efficiency
An original paper frame and two fabric layers mimic a filter stack on a tabletop. No filtration or performance test is performed.
| Record | Source | What it records |
|---|---|---|
| P1 | Assembly drawing | Arrow on frame marks the intended incoming face; coarse blue layer must face the arrow on this toy. |
| P2 | Front photo | Arrow and fine white layer are on the same face in sample X. |
| P3 | Back photo | Coarse blue layer visible on the opposite face of sample X. |
| P4 | Record status | No airflow or particle-retention measurements exist. |
Question for the model
What defect can be annotated from the photos?
Show the reference answer and checks
Annotate an orientation mismatch in sample X: P1 expects blue at the arrow face, while P2 shows white there and P3 locates blue on the opposite side. P4 means neither filtration performance nor the consequence of the reversal can be inferred.
- Using camera-left to infer the incoming face
- Inferring a performance loss without an actual measurement
Fields to preserve
- face_marker
- Anchors orientation when the sample is rotated
- visible_layer
- Records what the annotator can see on each face
- performance_evidence_status
- Keeps assembly evidence separate from testing
Test another case
Show a sample with an occluded arrow and otherwise clear layers; mark the orientation unresolved until the face can be identified.
Use inert fabric and invented drawings; no safety ratings, proprietary filters or testing instructions.
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.