herculaneum

Registered human ground-truth ink evaluation for the open Vesuvius Challenge SOTA scroll data

Same-winding constraints for the villa spiral fit, derived from published ink detections (PHerc. Paris 4). Preregistered A/B campaign in progress.

Byte-exact multiscale pyramid verification for Vesuvius Challenge open data. Read-only, no credentials.

Audited traced surface geometry on PHerc. MAN5, with the gate that decided what counted and an independent CT check

Text-line phase as an independent sheet-tracking signal for Herculaneum scrolls. Cuts sheet-identity errors 72%/44% on two scrolls, held-out.

Trace-quality reports for Vesuvius Challenge scroll segments — flags likely segmentation errors from an ink-detection prediction (CPU-only).

Diagnostics for Vesuvius Challenge surface-label datasets: split leakage, label topology, and how far labels sit from the ridge the scan actually shows.

Model-free physical ground truth for Vesuvius Challenge surface models: registers the two scroll pairs with a second higher-resolution public scan (PHerc0139 held-out 4.1um, Grand-Prize PHerc1203 2.4um), ships truth labels for both plus a standalone evaluator with shifted-null controls

Measure Before You Hunt — scan-quality + sheet-separability index for the 13 Grand-Prize scrolls, a validated ink instrument, and a corpus-wide screen (Vesuvius Challenge)

Certified topology auditing and transverse-clean tifxyz outputs for a five-scroll corpus. Report-only by default; finds where a traced surface passes through itself.