Synthetic South Korean Passport Images for OCR Model Training

More than 1,000 AI-generated passport images provide a privacy-safe resource for training computer vision models. The dataset is fully synthetic, containing no real personal data, and was created by ud-synthetic. It was last updated on May 14, 2026.

Use Cases

• Train optical character recognition (OCR) models based on synthetic passport text fields.
• Develop computer vision models for document layout and security feature detection based on AI-generated images.
• Benchmark identity document processing algorithms on a privacy-compliant dataset.
• Test model robustness against variations in synthetic passport designs and data placements.

Strengths

• Contains more than 1,000 synthetic images, providing a substantial volume for model training.
• All data is synthetically generated, ensuring no privacy concerns related to real individuals.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The synthetic nature may not capture all the nuances and imperfections of real-world documents.

Provenance

Source

ud-synthetic

Collection Method

AI-generated synthetic data.

Freshness

Last updated 2026-05-14 15:05:24; freshness should be verified.

Geography

South Korea (implied by document type)