CMC ranks among the top 12 globally in text recognition.

The CATI-VLM (Visual Document Understanding) model developed by the CMC Institute of Applied Technology (CMC ATI) has surpassed many international competitors, reaching the top 12 globally and top 1 in Vietnam in the ranking recently announced by the Robust Reading Competition (RRC) in June 2025 in the Document Visual Question Answering (DocVQA) category.

RRC's ranking in the DocVQA category, June 2025.

Amidst the rapid digital transformation and the adoption of artificial intelligence in Vietnam, OCR (Optical Character Recognition) technology is playing an increasingly important role in document digitization, business process automation, cost savings, and improved management efficiency. However, given the unique characteristics of the Vietnamese language, including its accents and handwriting, the recognition problem goes beyond simply 'reading' characters; it requires a model capable of comprehensively understanding the context.

Recently, CMC Institute of Applied Technology (CMC ATI) announced the CATI-VLM (Visual Document Understanding) model – developed by its research team from a large 5TB data warehouse – surpassing many international competitors to reach the top 12 globally and top 1 in Vietnam in the ranking published by Robust Reading Competition (RRC) in June 2025 in the Document Visual Question Answering (DocVQA) category.

The Robust Reading Competition (RRC) is a prestigious scientific competition (https://rrc.cvc.uab.es/) organized by the Computer Vision Centre (CVC) of the Autònoma de Barcelona University (UAB), Spain, a world-renowned research institution in the field of computer vision.

Initiated in 2011, the competition is held annually within the framework of the International Conference on Text Analysis and Recognition (ICDAR) – one of the world's leading forums in the field of computer vision. The competition attracts numerous researchers and engineers from universities, research institutes, and major technology corporations such as Tsinghua University, Hyundai Motor Group, and Tencent. The RRC problems are designed to promote technological progress, closely linked to practical problems ranging from translation and enterprise data management to urban analysis and historical document processing.

Dr. Dang Minh Tuan, Director of CMC ATI, shared: "We are delighted that the research capabilities of the CMC team have been affirmed through a prestigious global competition like RRC. In a short time, the research team has achieved a high ranking, demonstrating international competitiveness with major names from developed countries. More importantly, this is clear evidence of our ability to master technology to solve specific problems related to the Vietnamese language and specialized fields in Vietnam."

Dr. Dang Minh Tuan, Director of CMC ATI.

CATI-VLM differs from traditional OCR in that it not only extracts characters but also understands multiple layers of information: text content, non-text elements (tick boxes, checkboxes, charts, signatures, formulas), layout (page structure, tables, forms), and style (fonts, highlighting, etc.). The model can answer visual questions posed on document images, similar to ChatGPT, without needing to learn each specific form beforehand.

Notably, on the RRC ranking, CATI-VLM, with only 3 billion parameters, achieved the highest accuracy in 4 out of 7 datasets, outperforming many Big Tech models such as Deepseek (27 billion parameters), GPT-4 Vision Turbo + Amazon Textract OCR (top 34), and Baidu (top 22).

The achievement also demonstrates a practical approach, focusing on mastering core technologies and optimizing models to suit Vietnam's infrastructure conditions, rather than chasing after scalability and parameters.

Example of a university admissions application form

The text has been identified from the handwriting in the image above.

Mr. Nguyen Trung Chinh, Chairman of the Board and Executive Chairman of CMC Technology Group, emphasized: "This is the result of more than a decade of persistent investment in research and development (R&D) of technology. CMC's high achievements in the international technology arena affirm our strategy of mastering Vietnamese technology, coupled with our orientation towards AI transformation and expansion into the global market. We believe that Vietnamese intelligence is fully capable of competing with global Big Tech, creating a worthy position on the world technology map."

CATI-VLM will be applied in the C.OpenAI ecosystem of products, including: the CLS virtual assistant for reviewing legal documents, CMC SmartDoc - a digital document transformation platform, the CMC KMS knowledge management system, an automated reporting system for smart offices, and next-generation Agentic Documents applications.

QUANG HUY

Source: https://nhandan.vn/cmc-dat-top-12-the-gioi-ve-nhan-dang-van-ban-post891252.html