5CD-AI/Viet-Handwriting-OCR-v2

Dataset Overview

From our experience, OPEN-SOURCE DATASETS AND MODELS FROM THE GLOBAL COMMUNITY HAVE HELPED US GREATLY. However, we also learned that TO PUSH THE MODELS FURTHER, WE NEED MORE HIGH-QUALITY LOCAL DATA to DEVELOP STRONGER VIETNAMESE AI MODELS 💪.

This dataset consists of 60,248 Vietnamese 🇻🇳 handwritten text images collected and curated for Handwritten Text Recognition research .

The original images were crawled from public internet sources. For each image, only one bounding box containing handwritten text was selected. All samples were then manually annotated by human labelers, ensuring both high diversity and high transcription accuracy.

Compared to our v1, the v2 release adds 35,845 new training samples, significantly expanding the diversity and coverage of Vietnamese handwriting styles.

We are actively scaling to hundreds of thousands of HUMAN-VERIFIED samples using the same rigorous annotation process.

Dataset Structure

The dataset is released with two splits:

• Train: 59,248 samples
• Test: 1,000 samples

Examples

• Example 1
  

cô đã tạo ra nhiều hoạt động sân chơi cho các con,

• Example 2
  

Câu 2: Phát biểu quy luật phân li và phân li độc lập (4đ)

• Example 3
  

được chân thành cảm tạ các y, bác sĩ trong khoa, và đặc biệt 

• Example 4
  

Con cảm ơn bác sĩ Đức đã hỗ trợ tả thuốc Nụ cười cho các trường học.

• Example 5
  

Ngày 1/5/1951 Khai mạc đại hội chiến sỹ thi đua và cán bộ

• Example 6
  

- Đảm bảo tình hình ANTT, trong và ngoài khu vực

• Example 7
  

Vì số tuổi theo thời gian mỗi năm tăng lên 1, mà năm

• Example 8
  

Lặng thầm Thầy vẫn đưa đò qua sông

• Example 9
  

C2: Xác định cấu tạo của câu "Đẹp vô cùng tổ Quốc

• Example 10
  

- While được sử dụng để chỉ một khoảng thời gian dài trong quá khứ,

Data Notes & Terms

This dataset is compiled from publicly available sources on the Internet and has undergone a rigorous processing workflow to ensure compliance with current legal standards (Decree No. 13/2023/ND-CP on Personal Data Protection in Vietnam and international Fair Use principles):

• Data Anonymization: All images have been cropped into individual lines or single sentences. We employ both automated filtering and manual review processes to guarantee that the dataset does NOT contain any personally identifiable information (PII) (such as names, phone numbers, addresses, citizen ID numbers, etc.). This data is considered anonymized and is used solely for academic purposes and Vietnamese OCR/AI technology research.
• Non-Commercial Purpose: The dataset is released for free under a non-commercial open-source license CC BY-NC 4.0 and is strictly committed to not being used for any profit-generating activities.
• Takedown Policy: Despite our best efforts to anonymize the data, if you (the owner of the handwriting or the original content) identify any image belonging to you that you do not wish to be publicly available, please contact us via email at: huynhgiabaoa2@gmail.com or dtkhangbk@gmail.com. We pledge to review, verify, and remove the relevant images from the dataset within 24 to 48 hours.

Citation

@misc{doan2024vintern1befficientmultimodallarge,
      title={Vintern-1B: An Efficient Multimodal Large Language Model for Vietnamese}, 
      author={Khang T. Doan and Bao G. Huynh and Dung T. Hoang and Thuc D. Pham and Nhat H. Pham and Quan T. M. Nguyen and Bang Q. Vo and Suong N. Hoang},
      year={2024},
      eprint={2408.12480},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2408.12480}, 
}