Video OCR Evaluation Dataset for Multimodal LLMs

Designed to evaluate the Optical Character Recognition capabilities of Multimodal Large Language Models in video scenarios. It was created by DogNeverSleep for the associated research paper and was last updated in October 2025.

Use Cases

• Benchmarking OCR performance of Multimodal LLMs on video frames containing text.
• Evaluating video understanding models on tasks requiring text extraction from video scenarios.
• Training or fine-tuning models for text-in-video recognition tasks using the provided video samples.

Strengths

• Dataset is associated with a specific research paper (MME-VideoOCR), providing a clear academic context.
• Last update was recorded on 2025-10-10, indicating recent maintenance.

Limitations

• The dataset description explicitly prohibits commercial use, limiting its applicability.
• Key technical specifications like row count, column names, and file formats are unknown.
• The dataset is intended solely for academic research, restricting its user base.

Provenance

Source

Paper: MME-VideoOCR: Evaluating OCR-Based Capabilities of Multimodal LLMs in Video Scenarios.

Collection Method

Video samples are collected from a publicly available resource.

Freshness

2025-10-10