HeOCR/heocr-r4d-segmentation

HeOCR R4d: Hebrew document baseline segmentation (kraken)

R4d is a kraken baseline segmentation model for pages of modern Hebrew documents. It finds:

• each text line's baseline, with a stamp line type for lines inside stamps;
• the line polygons;
• eight region types: title, body, stamp, signature, footer, margin, identifier and table cells.

HeOCR trained it from scratch on synthetic pages only.

R4d is the segmentation step of HeOCR's three-model pipeline:

• R4d finds the lines;
• A1 puts them in reading order;
• R3h reads them.

The three are released together, as separate records that link to each other (see The pipeline).

It is a research baseline. All of its evaluation is on synthetic pages; no reviewed evaluation on real scans exists yet.

File	heocr-r4d-segmentation.safetensors (trained as best_0.8618.safetensors), 5,056,800 bytes
sha256	d32cd1a8efe7354445ccfea76551cc9943ca05792c332b5865f9bec985fe7bf7
Model type	segmentation (baselines and regions)
Format	kraken safetensors; trained and tested with kraken / ketos 7.0.2
Licence	Apache-2.0, Copyright (c) 2026 HeOCR maintainers (see LICENSE and NOTICE)
Training record	r4d_training_run.json (heocr_engine_training_run.v1)
Code	HeOCR/heocr-engine

The pipeline

Step	Model	Record
Segmentation: baselines, line polygons, typed regions	R4d, this model	10.5281/zenodo.24023070
Reading order: line order from each line's region type	A1, heocr-a1-reading-order.safetensors (69aa1946…)	10.5281/zenodo.24023072
Recognition	R3h, heocr-r3h-recognition.safetensors (f76affa8…)	10.5281/zenodo.24023074

A1 needs heocr-engine. kraken 7.0.2 cannot run A1 by itself, for two reasons:

• its neural reading order passes the segmenter's line types to A1, while A1 was trained on region types. kraken rejects the class-mapping mismatch;
• its text output ignores a learned line order (mittagessen/kraken#820).

Run the whole pipeline from a heocr-engine checkout (pip install '.[kraken]'):

heocr-engine kraken --api python --segmentation-model heocr-r4d-segmentation.safetensors --reading-order-model heocr-a1-reading-order.safetensors --reading-order-classes region --recognition-model heocr-r3h-recognition.safetensors page.jpg page.json

Architecture

R4d uses ketos 7.0.2's default segmentation network, passed explicitly and otherwise unchanged:

[1,1800,0,3 Cr7,7,64,2,2 Gn32 Cr3,3,128,2,2 Gn32 Cr3,3,128 Gn32 Cr3,3,256 Gn32 Cr3,3,256 Gn32 Lbx32 Lby32 Cr1,1,32 Gn32 Lby32 Lbx32]

Pages are scaled to a height of 1,800 pixels, in colour.

Classes:

• line types: default and stamp;
• regions: title, body, stamp_area, signature_area, footer, margin, identifier_area and table_cells.

These are hocrsyngen's region names.

Uses

• Baseline segmentation of modern Hebrew document pages: letters, notes, archive cards, ledgers, forms and notices, in print or handwriting-style fonts. It is best used with A1 for reading order and R3h for recognition.
• A research baseline for Hebrew document segmentation.

Out of scope

• Real scans, without checking first. There is no reviewed real-page evaluation.
• Layouts unlike hocrsyngen's seven templates. This includes multi-column books, newspapers and manuscripts.
• Reading order on its own. kraken's heuristic order is poor with R4d (see below), so use A1.

How to get started with the model

The full pipeline (R4d + A1 + R3h). This is the configuration the figures below come from. In a checkout of heocr-engine, run pip install '.[kraken]', then:

heocr-engine kraken --api python \
  --segmentation-model heocr-r4d-segmentation.safetensors \
  --reading-order-model heocr-a1-reading-order.safetensors --reading-order-classes region \
  --recognition-model heocr-r3h-recognition.safetensors \
  page.jpg page.json

Tested end to end with kraken 7.0.2 on CPU, one seed-101 page in about 8 s. The text matched the scored prediction exactly.

kraken alone (no A1; kraken's heuristic line order):

kraken -i page.jpg page.txt segment -bl -d horizontal-rl -i heocr-r4d-segmentation.safetensors \
  ocr -m heocr-r3h-recognition.safetensors --reorder --base-dir R

• kraken 7.0.2 cannot use A1, for two reasons:
  • kraken gives it the segmenter's line types (default, stamp) rather than region types, and rejects the pair;
  • kraken's text output ignores a learned order (mittagessen/kraken#820).
• kraken 7.0.2 sometimes aborts on exit ("recursive_mutex lock failed") after writing its output. Retrying works.

Training details

Training data

All synthetic. Nothing was scraped, and no real document or third-party text was used.

• Pages:
  • 1,800 pages from hocrsyngen at 8698895 (generator s10c-gold-complete-v1), seed 5101, over the seven templates. Region polygons come from the generator (S11a).
  • The split is 1,710 training pages (27,487 lines) and 90 validation pages (1,553 lines), assigned per page by hash.
• Fonts: Alef (1,286 pages) and Gveret Levin (514 pages), both SIL OFL 1.1. Credits are in NOTICE.
• Text:
  • The hocrsyngen corpus S10b (hebrew_lines_train_v1.txt, 439,596 bytes, sha256 a3f2f0db…).
  • It is original text authored by AI agents (Claude Opus 5.5 sessions) for HeOCR, with no third-party text, and dedicated to the public domain under CC0 1.0.

Training procedure and hyperparameters

• Initialisation: from scratch. No blla or other third-party weights were loaded.
• Trainer: ketos 7.0.2 segtrain through heocr-engine train-kraken-seg (heocr-engine aa67fa7).
• Optimiser:
  • AdamW, learning rate 0.0002, constant schedule, weight decay 1e-5;
  • baseline width 8, baseline tolerance 10, dice weight 0.5, no augmentation.
• Stopping: early stopping with lag 8 (minimum 10 epochs). It stopped after 29 epochs; the best was epoch 20.
• Validation: mean IoU 0.8618. Baseline precision, recall and F1 at the best epoch were 0.9994, 0.9800 and 0.9890.
• Run: seed 42, on Apple mps. Not deterministic.

Evaluation

Testing data, factors and metrics

The batches are those of the R3h card. None was trained or tuned on, apart from the hocrgen fixture, a development set. All gold is synthetic.

Line geometry

Polygon IoU ≥ 0.5 against gold lines (heocr-engine diagnose-segmentation). blla is kraken's default segmenter, with -d horizontal-rl.

Segmenter	Batch	P / R / F1	F1, baseline match	Gold lines missed	Gold lines split	Kendall τ of line order
blla	seed-101	0.733 / 0.800 / 0.765	0.842	27	78	0.838
R4d, kraken's heuristic order	seed-101	0.913 / 0.914 / 0.913	0.995	1	19	0.819
R4d + A1 order	seed-101	0.913 / 0.914 / 0.913	0.995	1	19	1.000
blla	held-out	0.624 / 0.726 / 0.671	0.773	28	96	0.811
R4d, kraken's heuristic order	held-out	0.915 / 0.916 / 0.915	0.998	0	4	0.865
R4d + A1 order	held-out	0.915 / 0.916 / 0.915	0.998	0	4	1.000

Stamps: R4d matches 25 of 40 stamp lines on seed-101 and 35 of 40 on held-out.

End to end (R4d + A1 + R3h)

CER under hocrgen's primary.v1 profile:

Batch	R4d + A1 + R3h	blla + R3h
seed-101	0.020	0.168
held-out	0.010	0.238
eval-mixed-heldout	0.017	0.230
eval-typo-heldout	0.009	0.291
eval-layout-heldout	0.124	0.498
hocrgen fixture (development set)	0.130	0.404

Bias, risks and limitations

• The synthetic gap. The evaluation pages share training's fonts, templates and degradations, so these scores overstate accuracy on real scans.
• Split and merged lines:
  • On seed-101, 19 gold lines are split, and 14 predicted lines each merge two gold lines.
  • Some stamp lines are merged into the body lines they overlap: 8 on seed-101.
• Line order without A1 is poor. kraken's heuristic reads separated stamp lines before the body lines around them (Kendall τ 0.819 on seed-101; 9 of 80 pages fully in order).
• Ledgers:
  • R4d's training pages predate hocrsyngen's S11b ledger layout, so newer ledger pages (the fixture, eval-layout-heldout) look unlike its training.
  • eval-layout-heldout's ledger CER is mostly hocrgen's | gold convention (HeOCR/hocrgen#98).
• kraken's polygonizer sometimes reports "failed on line 0" on a page. The page still completes.
• Layouts: only hocrsyngen's seven templates were seen in training.

Licence and attribution

• Licence: Apache-2.0. Copyright (c) 2026 HeOCR maintainers.
• NOTICE: redistributions must keep the NOTICE file. It credits:
  • kraken (Apache-2.0);
  • the fonts Alef (HaGilda & Mushon Zer-Aviv, OFL-1.1) and Gveret Levin (The Gveret Levin Project Authors, OFL-1.1);
  • the CC0 synthetic training text.
• Fonts: the model contains no font data, and the fonts are not distributed with it.

Provenance and reproduction

Part	Identity
Training batch	hocrsyngen 8698895, seed 5101; manifest 29dffcca283770c56dedb4e53b57ccae289ec37124fd4399d968d93a3dd9295a; line annotations 871355fdecb36fd561c1ee9832d8b6aec221685cca6ff3f13cad6b60cd6f36bf
Training text	a3f2f0db1bf48d8cdd618579af42ca36710cadf81cdc4d5e89d2dc6c192e6611 (hocrsyngen hebrew_lines_train_v1.txt)
Split	per page by hash, 5% validation; assignment a794ebf3dbe5dd322cc85197a133207409fea0e487d02a23a52b0c76886d4006
Training code	heocr-engine aa67fa7, ketos 7.0.2
Training record	r4d_training_run.json (heocr_engine_training_run.v1), in this deposit
Scoring	hocrgen 2d3efb0 and c8fb74e; heocr-engine 55b0f7d (end to end), 129c58e and eed1839 (line geometry)

Contact

Report problems and questions as issues on HeOCR/heocr-engine.

Citation

Cite this model by its version DOI, 10.5281/zenodo.24023070. Never cite the concept DOI.