Glossary · 28 Sep 2026

What is OCR?

← Glossary

OCR, or optical character recognition, is technology that turns images of text, from scans, photos or PDFs, into text a computer can read, search and use. IBM defines it as "a technology that uses automated data extraction to quickly convert images of text into a machine-readable format". Newer systems use AI, which IBM notes can extend recognition to "identifying languages or handwriting".

Print and handwriting

Print reads well. A benchmark published in August 2026 notes that the top model on a separate printed-document benchmark, OmniDocBench, reaches 96.34%. On its own set of 500 real handwritten documents, mostly Chinese and English, the best current AI model scored 71.85 on a composite of text, table and formula measures. People transcribing the same pages in one pass scored 77.09, a gap the authors call "narrow".

Tables were the weak point. The best table score "indicates nearly 40% structural and content mismatch". And "63-91% of model errors are prior-driven versus only 49% for humans": the models more often wrote something plausible that wasn't on the page.

In a timber business

Timber paperwork mixes print and handwriting, and much of the handwriting sits in tables of numbers. OCR alone turns the page into text. Turning that text into records, with the right number in the right field, is document extraction.

From Quarri's own work with a forestry operation: a pilot turned 20 handwritten trip tickets into structured load records. A pilot of that size shows it can be done. It doesn't yet show an error rate.

What it isn't

OCR isn't document understanding. It gives you the words, not which word is the volume. Nor is it a guarantee: the output needs checks, especially for numbers.

What to check

Test on your own documents, including the worst ones. Compare the error rate with manual keying of the same pages, measured by field. Check handwritten figures against another record of the same event, such as a weighbridge weight.

Quarri for operations teams is built for the people running the crews, the lines and the yard.

Sources

  1. IBM, "What is OCR?", 18 April 2024: ibm.com
  2. Zhang and others, "WildHandBench: A Benchmark for Handwritten Text Understanding that Challenges MLLMs and Humans", arXiv 2608.22959, 24 August 2026: arxiv.org
  3. Quarri evidence ledger, E23 (pilot)

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.

See it on your own data.

Live in two weeks, on the systems you already run.