OmniHandwritingOCR

This is the official dataset repository accompanying the paper:

OmniHandwritingOCR: A Diagnostic Benchmark for Evaluating Multimodal LLMs in Handwritten OCR Scenarios

Accepted at CIKM 2026 (full research papers track).

Status

Data release pending institutional review.

This repository currently contains documentation only. No dataset files are publicly available at this stage.

Subject to applicable institutional authorization, privacy, rights, and ethical review, we intend to release only the eligible and de-identified portion of our newly collected private dataset, together with its annotations, data splits, and evaluation resources.

Samples without sufficient authorization or reliable de-identification will remain non-public.

Third-Party Public Datasets

Our benchmark also evaluates models on several existing public datasets.

We will not redistribute, mirror, or repackage the images, annotations, or other data files from these third-party datasets. Users should obtain them from their original providers and comply with their respective licenses and terms of use.

Availability

Coming soon, subject to the completion of the required authorization, privacy, and rights reviews.

Paper

Paper link: Coming soon.

Contact

For questions regarding this dataset, please contact the authors of the paper.