{
  "title": "Potato bruise and cut-sample images for AI",
  "source_page": "https://fiund.com/industries/potato-bruise-cut-sample-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "A surface streak and an interior mark require specimen-matched views",
    "context": "Original textual annotations describe fictitious training tubers and already authorized view types. No cutting, product damage, actual images or official grade decision is performed.",
    "records": [
      [
        "PT1",
        "Tuber A paired views",
        "Exterior has a dark streak. Reviewer labels it superficial soil; matching authorized interior-view annotation shows no mark in its reviewed region."
      ],
      [
        "PT2",
        "Tuber B paired views",
        "Exterior reviewed as visually clear in the supplied face. Matching interior-view annotation records an internal mark at region R."
      ],
      [
        "PT3",
        "Tuber C",
        "Exterior-only view with a dark region. No interior view or reviewed explanation supplied."
      ]
    ],
    "question": "Which tuber has a documented interior mark, and can C's exterior appearance resolve its interior condition?",
    "answer": "PT2 supplies the interior mark for B despite its clear-looking exterior face. PT1 separately documents A's superficial soil label and the limited interior region reviewed. PT3 lacks an interior view and an explanation for its dark region, so C's interior condition remains unresolved. Preserve specimen-matched view and reviewer scope instead of using skin appearance as proof of internal condition or borrowing another tuber's annotation.",
    "failureModes": [
      "Predicting an interior finding from an exterior mark alone.",
      "Pairing a cut-surface view with the wrong specimen or generalizing beyond its reviewed region."
    ],
    "fields": [
      [
        "Tuber identity and paired-view link",
        "Preserves specimen continuity."
      ],
      [
        "Exterior versus interior region",
        "Separates observation types."
      ],
      [
        "Reviewer explanation and missing view",
        "Makes uncertainty explicit."
      ]
    ],
    "holdout": "Use a different authorized skin variety with a naturally darker exterior. Require direct reviewed evidence rather than a learned color shortcut.",
    "boundary": "Use owned approved samples and original annotations. Clear grading-station imagery separately; exclude rejects, field IDs, brands and buyer terms."
  }
}