GlotOCR-bench: 16,375 Images Covering 158 Writing Systems

GlotOCR-bench is a dataset of 16,375 images covering 158 writing systems and over 2,000 languages, designed to evaluate fundamental OCR capabilities. It was created by cis-lmu and was last updated on the platform on April 15, 2026. The dataset is released under the GlotOCR Open Evaluation License v1.0.

Use Cases

• Benchmarking OCR model performance across diverse writing systems based on the 158 systems covered.
• Evaluating language support in OCR tools based on the over 2,000 languages included.
• Training or fine-tuning OCR models for low-resource scripts based on the dataset's multilingual scope.

Strengths

• Contains 16,375 images for evaluation.
• Covers a wide scope of 158 distinct writing systems.
• Includes text from over 2,000 languages.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or source bias inherent to the collection method.

Provenance

Source

cis-lmu

Collection Method

Images were generated from texts; the specific generation method is not detailed in the provided input.

Freshness

Last updated 2026-04-15 18:46:40; freshness should be verified.