thuongdz/ocr_annual_financials

TiniX越南OCR年度财务报表数据集是一个包含越南上市公司2015年至2025年年度财务报表OCR文本的数据集。数据由TiniX AI收集和处理，包括18,231份越南语报告，对应1,491个不同的股票代码。每份文件包含一份财务报表的完整OCR内容，涵盖合并财务报表、母公司财务报表、年度审计报告、财务报表附注及相关附录和表格。OCR内容保持了原始文本结构，数字和表格数据的准确率达到95%。数据集按股票代码、报告年份和报告名称分层组织，以TXT格式存储。该数据集适用于OCR基准测试、金融视觉语言模型训练、越南语金融NLP、检索增强生成系统、财务信息提取以及大型语言模型预训练等领域。The TiniX Vietnam OCR Annual Financial Statements dataset is a collection of OCR text from annual financial reports of listed companies in Vietnam from 2015 to 2025. The data is collected and processed by TiniX AI, comprising 18,231 Vietnamese reports corresponding to 1,491 different stock tickers. Each file contains the full OCR content of a financial statement, including consolidated financial statements, parent company financial statements, annual audit reports, financial statement notes, and related appendices and forms. The OCR content preserves the original text structure, with an accuracy rate of 95% for numerical and tabular data. The dataset is organized hierarchically by stock ticker, report year, and report name, stored in TXT format. It is suitable for OCR benchmarking, financial vision language model training, Vietnamese financial NLP, retrieval-augmented generation systems, financial information extraction, and large language model pretraining.