extract-images

To extract illustrations from historical book scans, this Skill detects bounding boxes, classifies types like woodcuts or maps, and generates structured metadata with scholarly descriptions. The end-to-end workflow spans OCR, extraction, metadata generation, and gallery curation.

Metadata for historical book illustrations includes bounding box coordinates, type classifications like emblem or frontispiece, gallery quality scores from 0.0 to 1.0, and structured fields covering subjects, figures, symbols, style, and technique.

Yes, the Skill provides API-friendly endpoints for triggering illustration extraction from specified books and listing generated gallery results. This supports automated digitization workflows across a library collection.

Automating image detection and classification for digitization projects is best handled by an end-to-end workflow that applies AI vision to detect illustrations, classify types like engravings or diagrams, and produce museum-style descriptions with quality scoring.

Yes, the image extraction tool supports bounding box coordinates for each detected illustration and gallery quality scoring on a 0.0 to 1.0 scale, alongside museum-style descriptions to support curation and research workflows.