Ink2Index

Newspaper article extraction

Newspaper processing workspace

Turn a newspaper scan into structured articles.

Upload a newspaper scan, run the article pipeline, and read reconstructed articles with extracted metadata from one workspace.

Use a multipage .tif/.tiff or a single-page .png/.jpg/.jpeg.

Choose a newspaper image Click to browse from your computer

Newspaper title Publication date

The available pipeline runs with the project defaults. Fine-grained controls remain in Diagnosis.

Detect layout blocksYOLO document layout model

Available

Group article blocksPairwise scoring and clustering

Available

Reconstruct article textReading order and OCR text assembly

Available

Extract metadataSmall-parameter Qwen LLM

Available

Upload a newspaper to begin.

Document upload, processing, review, and export.

Newspaper image

Source Publication date

Idle

YOLO layout detection Pairwise model Article grouping OCR features Cluster validation (required) Reuse YOLO cache

YOLO confidence 0.25

Image size Same-column K Adjacent-column K Cross-column K Model version Pairwise threshold Clustering Leiden resolution

No job started.

Pages0

Detections0

Articles0

Pairs0