Submit a Paper

Help us track the state of the art. Submit papers with benchmark results.

What we're looking for

- Papers with benchmark results on OCR datasets (ICDAR, SROIE, OmniDocBench, etc.)
- New models or methods that advance the state of the art
- Papers introducing new benchmarks or datasets
- Code availability is a plus (we prioritize reproducible results)

Paper Title *

arXiv URL

Preferred. We'll extract the arXiv ID automatically.

Alternative Paper URL

If not on arXiv (OpenReview, ACL Anthology, etc.)

Code Repository

Research Area

Datasets Evaluated On

Select all that apply

Benchmark Results (Optional)

Notes

Your Email (optional)

We'll notify you when the paper is added

Our database tracks the relationships between papers, benchmarks, datasets, and models:

Research papers introducing models, methods, or benchmarks

Evaluation benchmarks like ICDAR, SROIE, OmniDocBench

ML models evaluated on benchmarks (PaddleOCR, TrOCR, etc.)

Benchmark scores linking models to datasets