Audit unanimous OCR labels for completeness and association

Audit a sample of unanimous OCR labels against the original pixels for identifier completeness and object association, not only character agreement. Independent models can agree on the same visible fragment while failing to establish whether hardware hides another character, or whether a leading mark is branding rather than an identifier prefix. Route those cases to unresolved adjudication instead of promoting agreement to ground truth. Keep raw transcriptions and semantic eligibility separate so a normalization rule cannot erase the disagreement. This applies to model-derived supervision; it does not establish human-certified error rates or imply that every partially obstructed image is unreadable.

Applies when: accepting whole-image identifier labels from multiple vision models. Keywords: OCR annotation agreement completeness association occlusion logo prefix normalization