Details

Autor(en) / Beteiligte

• Nguyen, Anh Thi Lan
• Do, Quan Hong
• Doan, Tung Trung
• Vu, Vu Viet
• Le, Cuong

Titel

FGW-VnText: A Deep Learning-based System for Detecting and Recognizing Vietnamese Identity Cards and Citizen Identities

Ist Teil von

• Proceedings of the 2023 8th International Conference on Intelligent Information Technology, 2023, p.145-151

Ort / Verlag

New York, NY, USA: ACM

Erscheinungsjahr

2023

Quelle

ACM Digital Library

Beschreibungen/Notizen

• Electronic Know Your Customer (eKYC) technology is an online solution to identify and verify a customer’s identity, which is now widely used in the banking sector or public administrative services. One of the first critical steps in any eKYC process is extracting identity cards (ID Cards) or citizen identification (Citizen ID) information. These identity documents, however, are often varied in even each country. For example, in Vietnam, three valid but different types exist: 12-digit ID Card, 12-digit Citizen ID without a chip, and 12-digit Citizen ID with a chip. Intending to build an effective and robust end-to-end system for detecting and automatically extracting information from different types of ID Cards and Citizen IDs, this study proposes a system that integrates YOLOv5, CRAFT detection, and VietOCR in one deep learning pipeline. The experimental results on the collected data show that our proposed model brings more accurate performance than some recent related studies. The exact match results on the 12-digit ID card reached an average of 94.64%, the 12-digit citizen ID without a chip reached an average of 93.15%, and on the 12-digit citizen ID with a chip, an average of 90.57%.