The Dataset Viewer has been disabled on this dataset.

KoOCR-Bench v2.0c

Korean document OCR benchmark: tables, official gazette, statistics, KDI reports, user documents, 2026 CSAT (์ˆ˜๋Šฅ), handwriting, accounting disclosures, drug labels, patents, medical forms. 12 slices, 877 public pages + 193 held out (1,070 in all). v2.0c (2026-10-29): handwritten_rw11 is reduced to the 111 pages whose truth was reviewed by hand.

Not for training: every page here is eval-only.

Slices

slice pages (this repo) held out license / source
gwanbo_table 81 19 ๊ด€๋ณด (public official gazette)
handwritten_owner20 18 2 ๊ตญ๋ฆฝํ•œ๊ธ€๋ฐ•๋ฌผ๊ด€ ์†Œ์žฅ์ž๋ฃŒ (rights per source); ๊ต๋ณด ์†๊ธ€์”จ (rights per source); realworld photo (owner-provided); AIHub 81 (NIA AI Hub; rights per AIHub terms)
handwritten_rw11 90 21 ๊ตญ๋ฆฝํ•œ๊ธ€๋ฐ•๋ฌผ๊ด€ ์†Œ์žฅ์ž๋ฃŒ (rights per source); ๊ต๋ณด ์†๊ธ€์”จ (rights per source); realworld photo (owner-provided); AIHub 81 (NIA AI Hub; rights per AIHub terms)
kdi_headings 78 22 KDI (owner-cleared for public release 2026-09-30)
kostat_wide 80 20 KOGL (per-article type) โ€” 1์œ ํ˜• (์ถœ์ฒ˜ํ‘œ์‹œ), logo in attachment
suneung2026 87 17 KICE 2026 ์ˆ˜๋Šฅ ๋ฌธ์ œ์ง€ (owner-provided, eval only)
tables 86 14 KOGL 1์œ ํ˜• (๊ณต๊ณต๋ˆ„๋ฆฌ ์ œ1์œ ํ˜•: ์ถœ์ฒ˜ํ‘œ์‹œ)
user_real 58 17 ggugeo ToS (owner-cleared for public release 2026-09-30)
accounting_table 87 13 DART ๊ณต์‹œ (public disclosure)
drug_label 80 20 MFDS ์˜์•ฝํ’ˆ์•ˆ์ „๋‚˜๋ผ ํ—ˆ๊ฐ€์ •๋ณด (owner-cleared 2026-10-27)
patent_exp 85 15 KIPO ํŠนํ—ˆ๊ณต๋ณด (published gazette; owner-cleared 2026-10-27)
medical_form_synth 47 13 synthetic (generated, no real patient data)
total 877 193

Suneung: owner reports KICE OK for this release (2026-10-27).

handwritten_owner20 is a 20-page subset of the handwriting pages with owner-corrected truth (same images as rw11).

Total (v3 weights, 2026-10-31)

bash score.sh A=a.jsonl | python3 scorer/koocr_total_v3.py prints one weighted total per engine (higher is better; table slices use teds, text slices 1 - cer). v3 raises tables and handwriting; slice ratios inside each group are the v2 ones.

group slices (weight) sum
tables kostat_wide .159 ยท tables .136 ยท gwanbo_table .136 (scored by CER) ยท accounting_table .068 .50
text suneung2026 .056 ยท kdi_headings .044 ยท user_real .044 ยท patent_exp .022 ยท drug_label .022 ยท medical_form_synth .011 .20
handwriting handwritten_owner20 .10 ยท handwritten_rw11 .10 .20
reading order page-weighted mean of order over every slice that reports it .10

A slice an engine has no rows for is dropped and the remaining weights are renormalised (row marked partial). The v2 weights (tables .44 ยท text .36 ยท handwriting .10 ยท order .10) stay available as scorer/koocr_total_v2.py.

Results (v3 total, 2026-11-01; CLOVA and kordoc added 2026-11-09; LightOnOCR-3 added 2026-11-11; PepperOCR-VL added 2026-11-12; Gemini 4 Hydrogen, kordoc + HTML tables and the current Schift OCR 2 added 2026-11-13; Schift OCR 2 updated 2026-11-18)

Scored on all 1,070 pages (public + holdout). Higher is better. tbl_form / tbl_text split the grid-table slices into grid shape and cell text; they are shown, not added to the total.

engine total tables text hand order tbl_form tbl_text
Claude Opus 5.5 (claude-opus-5-5, API) 0.924 0.920 0.901 0.948 0.941 0.900 0.902
Schift OCR 2 (API) 0.919 0.920 0.906 0.927 0.932 0.910 0.908
OpenAI Codex (gpt-6-sol, API) 0.918 0.913 0.897 0.945 0.934 0.895 0.889
Google Gemini 4 Hydrogen (gemini-4-hydrogen, thinking low, via Antigravity CLI) 0.918 0.911 0.893 0.954 0.933 0.893 0.893
kordoc 4.24.57 + Markdown tables converted to HTML (local pipeline) 0.901 0.910 0.856 0.917 0.915 0.907 0.896
Schift OCR 1 beta 0.862 0.862 0.869 0.823 0.928 0.873 0.842
datalab-to/chandra-ocr-2 0.860 0.886 0.804 0.816 0.933 0.870 0.850
NAVER CLOVA OCR (General + table extraction, API) 0.853 0.874 0.736 0.943 0.800 0.884 0.875
kordoc 4.24.57 (GitHub 1d3dab1, local pipeline) 0.853 0.814 0.856 0.917 0.915 0.774 0.766
Qwen/Qwen3.8-27B-FP8 0.852 0.883 0.862 0.724 0.931 0.868 0.850
PaddlePaddle/PaddleOCR-VL-1.6 0.851 0.879 0.814 0.755 0.976 0.872 0.864
sionic-ai/PepperOCR-VL 0.810 0.751 0.849 0.836 0.976 0.726 0.673
lightonai/LightOnOCR-3-0.8B 0.776 0.848 0.861 0.435 0.928 0.845 0.795
MinerU 2.5 (local pipeline) 0.758 0.804 0.810 0.478 0.984 0.759 0.750
OpenGVLab/InternVL3_5-38B 0.753 0.753 0.688 0.738 0.910 0.797 0.704
mistralai/Mistral-Small-3.2-24B-Instruct-2506 0.696 0.756 0.676 0.435 0.962 0.772 0.678
google/gemma-4-26B-A4B-it 0.689 0.686 0.612 0.656 0.919 0.753 0.612

Schift OCR rows are our models (weights not released). Schift OCR 2 is the engine served since 2026-11-18 (earlier engines: 0.916 from 2026-11-13, 0.899 from 2026-11-01); pages below 1.2 MP are upscaled to 1.2 MP before it reads them, as the API does for every Markdown request.

kordoc + Markdown tables converted to HTML: the same kordoc 4.24.57 output as the kordoc row, with each Markdown table rewritten as an HTML <table> (the scorer reads tables only as HTML). Only the table slices change; every other slice scores the same as the kordoc row.

NAVER CLOVA OCR: General OCR V2 API with enableTableDetection: true, lang: ko, run 2026-11-09, score only (commercial API, predictions not published). Output is the API's words in API order (lineBreak ends a line); each detected table is written once as HTML <table> (colspan/rowspan from the API) where its first word falls, and words inside a table box are not repeated. Three handwriting photos over the API's 8,000 px limit were shrunk to 8,000 px on the long side. Its low text comes mostly from suneung2026: the API returns two-column exam pages as one large table, so the two columns interleave.

Google Gemini 4 Hydrogen: run 2026-11-13 through Google's Antigravity CLI (agy 1.5.6/1.5.7, print mode, --model gemini-4-hydrogen-low, thinking low), one call per page with the page image as a file and one fixed prompt for every page (transcribe as Markdown, tables as HTML <table>, math in LaTeX). The CLI has no temperature setting, so its default sampling was used; one run, with up to 3 tries only when a call failed. Score only (predictions not published). 3 pages (tables, gwanbo_table and drug_label, one each) were blocked by Gemini's content safety filter on all 3 tries and are scored as empty. Its hand (0.954) is the highest on the board (CER owner20 0.057, rw11 0.036); its text is lower mostly on suneung2026 (CER 0.187).

sionic-ai/PepperOCR-VL (AGPL-3.0, fine-tune of Qwen/Qwen3.5-4B): run 2026-11-12 with the repo's prompt.txt verbatim, processor default image size, thinking off, temperature 0, top_p 1, repetition_penalty 1.0, max_tokens 8192, one run. The card's defaults (repetition_penalty 1.05, no_repeat_ngram_size 30, retries at a higher temperature) were not used, so it runs under the same plain greedy decoding as every other engine here. 47 printed pages stopped at the 8,192-token cap; no handwriting page did. Its low tables comes mostly from kostat_wide (0.598) and tables (0.685); its order (0.976) is among the highest on the board.

lightonai/LightOnOCR-3-0.8B (Apache-2.0): run 2026-11-11 with the model card's image-only prompt (no text instruction), longest side resized to 2048 px, thinking off, temperature 0 (the card suggests 0.2), repetition_penalty 1.0, max_tokens 8192. 27 pages (7 printed, 20 handwriting) stopped at the 8,192-token cap. Its low hand is handwriting CER (owner20 0.549, rw11 0.581).

kordoc: open-source npm package by chrisryugj (not ours), GitHub HEAD 1d3dab1 (2 commits after npm 4.24.57), built from source and run on page images with default options (--jobs), output markdown with HTML tables, run 2026-11-09. Its CHANGELOG reports tuning against the public 877 pages; the score here is on all 1,070 pages, including the 193-page holdout it never saw.

Holdout

20% of every slice is held out and is not in this repo: a page is held out when int(sha1(id), 16) % 5 == 0 (for owner20 the hash is taken on its rw11 id, so a page is in the same split in both slices). Only counts are published.

Files

  • manifest.jsonl โ€” one row per page: id, slice, tier, image, truth, source, license (+ math_text_only, gt, split). truth.type is table_grid (rows of cells with text, cs, rs), page_text (text = path under truth/), or headings+page_text (adds headings = [[level, title], ...]).
  • images/, truth/
  • scorer/ โ€” KoOCR-Bench v1.3r scorer (koocr_score_v13r.py), GriTS on rendered grids (grits_render.py, unmodified microsoft/table-transformer grits.py, MIT), TEDS from opendataloader-bench (odlb_src/).

Scoring

pip install -r requirements.txt
bash score.sh MYENGINE=preds.jsonl      # preds rows: {"id": "<manifest id>", "text": "<engine output>"}

Output is format-neutral: Markdown syntax and HTML tags are stripped before text metrics; tables must be HTML <table>. Per slice: tables/kostat_wide โ†’ GriTS-based teds (0.5 form + 0.5 content) plus TEDS/TEDS-S/grid_exact; text slices โ†’ cer (pred blocks re-ordered to truth order), order, foreign; suneung โ†’ mark_keep; kdi โ†’ heading_recall. Per-page metrics go to scores.pages.jsonl.

Downloads last month
646