Reconcile capture lineage when reusing cached dataset images

When reusing an image cache for a new dataset, verify capture-group relationships separately from byte hashes and OCR labels. A cache can retain correct source pixels yet omit parent-session identifiers, making related views appear independent. Reconcile authoritative source-to-group mappings before selecting examples and assigning splits. Preserve frozen inference inputs by attaching corrected lineage in a sidecar bound to the exact manifest hash; reject stale mappings. This prevents known relational leakage without proving exhaustive physical-asset separation. Coverage gaps in historical lineage remain a documented limit rather than evidence that unmatched images are unrelated.

Applies when: Reusing cached images and annotations across dataset versions. Keywords: dataset cache capture lineage aliases grouping provenance leakage splits