Project's information

Project's title Developing deep learning methods for application in document analysis and recognition
Project’s code VAST01.01/19-20
Research hosting institution Institute of Information Technology
Project leader’s name PhD. Nguyen Duc Dung
Project duration 01/01/2019 - 31/12/2021
Project’s budget 600 million VND
Classify Fair
Goal and objectives of the project

- Develop new text structure recognition algorithm, including physical structure (position) and logical structure (format) based on deep learning approaches and existing research results. The new algorithm focuses on minimizing the detection error and the error of  discriminating/misidentifying the data blocks in the document text image page.
- Developing algorithms to improve image quality such as deblurring, image restoration, and super-resolution based on convolutional neural networks (CNN) to improve object recognition quality. image image.
- Developing algorithms to improve the computational speed of deep learning networks in image object processing and recognition. New algorithms ensure computing speed on mainstream computers (not equipped with GPUs) or mobile devices.

Main results Theoretical results: 01 Submit a scientific article in the journal International SCI / SCIE on methods to detect and analyze the structure of the table, locate the form, content identification and table form and 01 Scientidic article in  conference proceedings with the International criticism or national workshop.

Applied results: The program: document analysis and recognition.
For training: supporting 01 Phd student

Novelty and actuality and scientific meaningfulness of the results

We present TableSegNet, a compact architecture of a fully convolutional network to detect and separate tables simultaneously.

Products of the project

- Scientific papers in referred journals (list):

  • Duc-Dung Nguyen. “TableSegNet: a Fully Convolutional Network for Table Detection and Segmentation in Document Images”.  International Journal on Document Analysis and Recognition (IJDAR), 2021(SCIE)
  • Nguyễn Thị Thanh Nga, Nguyễn Đức Dũng, Lại Quốc Anh. Nghiên cứu ứng dụng mạng học sâu trong phân tích cấu trúc trang ảnh văn bản. Hội thảo quốc gia lần thứ XXII: Một số vấn đề chọn lọc của Công nghệ thông tin và truyền thông – Thái Bình, 28-29/6/2019, (tr 205-210)

- Technological products (describe in details: technical characteristics, place):

  • Overview Reports  and scientific papers
  • The program is analysis and segment tables.
Images of project
1655277431216-39. nguyen duc dung.png