Yuliang-Liu

AWESOME-OCR-LLM: Track OCR Research in the LLM Era

AWESOME-OCR-LLM is a curated reading list for research on OCR with large vision-language and multimodal models. It organizes papers on document parsing, understanding, visual text generation, and evaluation, with a focus on recent work.

View on GitHub Compare
Stars
747
Forks
56
Stars, last 30 days
+14
Last commit
2 days ago
Contributors
13

Our take

Activeregular commits

Health score

42/100

How it is scored
Activity
40
Community
59
Issues
21
Pull requests
46

+14 stars in the last 30 days

This is a useful, actively maintained research index, but its value is in discovery rather than implementation or standardized model comparison.

Good fit if

  • You need a focused reading list for OCR and multimodal document research.
  • You want papers grouped by task, with code or model links included when available.
  • You are comfortable assessing papers individually, since the repository warns that reported evaluations are not reliably comparable.

Look elsewhere if

  • You need a ready-to-run OCR tool, library, or hosted service.
  • You require a unified, reproducible benchmark or comparable leaderboard results.
  • You need a repository with releases or a tracked issue-resolution history as part of your adoption criteria.
All health signals
Last commit2026-11-11 (2 days ago)
Commits, last 90 days47
Releases, last 12 months0
Contributors13 (top contributor: 38% of commits)
Issues closed, last 90 days0
Pull requests merged, last 90 days2
Project age9 months

Checked on 2026-11-14 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

747

Forks

56

Change since Nov 16, 2026: 0 stars·0 forks

Overview

AWESOME-OCR-LLM is a curated, continuously updated research reading list about OCR in the era of large language models. It helps researchers and practitioners follow work that applies vision-language or multimodal models to text-rich images and documents, rather than trying to cover general document AI or classical OCR broadly.

The list spans document parsing and understanding, visual text generation, specialized models, and benchmarks. Entries link to papers and, where available, code or model resources. The repository is a literature guide, not an OCR implementation or an executable evaluation platform.

Key Features

  • Groups research into document parsing, document understanding, visual text generation, specialized models, and benchmarks.
  • Maintains dated paper tables with titles, affiliations, venues, and links.
  • Highlights daily papers and emerging research trends.
  • Provides separate full lists for document parsing and document understanding.
  • Includes guidance for contributors on ordering and metadata to include.
  • Notes that reported benchmark results may rely on self-reported numbers, private test sets, or inconsistent protocols.

Use Cases

  • Researchers entering OCR and document AI can use the categories to find relevant recent papers and research directions.
  • Teams comparing document-parsing approaches can scan linked papers and implementations before selecting candidates to evaluate.
  • Developers working on document question answering or multimodal retrieval can discover research on visual evidence, long documents, and document RAG.
  • Researchers studying text rendering in generated images can use the visual text generation section to find work related to OCR's generative counterpart.

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Browse the README and its linked full lists. No installation is required.

Considerations

  • The repository is young but active, with regular recent commits and contributions from multiple people. It has no releases, so it is maintained as a changing resource rather than a versioned product.
  • The README itself cautions that many reported results use technical reports, private test sets, and inconsistent protocols. Treat listed scores as reported, not as a rigorous cross-paper ranking.
  • Coverage depends on community curation. The README welcomes pull requests and asks contributors to provide venue or date, affiliation, and a code or model link.
  • The repository does not provide an OCR runtime, installation instructions, or a unified benchmark implementation. Use the linked papers and projects for practical evaluation.

Found this useful?

Share it with someone who would like AWESOME-OCR-LLM.