Professor Shuangping Huang guided Her Students to Win the Second Place in the 4th CSIG Image Graphics Technology Challenge Classification Competition

The 4th CSIG Image and Graphics Technology Challenge started on May 6, 2023 and ended on October 24, 2023. In this competition, the team composed of Lei Hu, Junwen Tan, Fan Yang and Ling Deng from Guangdong Branch of China United Network Communications Co., LTD., under the guidance of Professor Shuangping Huang, won the second place in the Table Structure Recognition Challenge after fierce competition in the preliminary, semi-final and final defense.

The CSIG Image Graphics Technology Challenge is a series of competitions organized by the Chinese Society of Image and Graphics (CSIG).The Table Structure Recognition Challenge is hosted by the Chinese Society of Image and Graphics (CSIG) and is organized by the National Engineering Research Center for Speech and Language Information Processing and IFlytek Co., LTD.It aims to promote the development and application of image graphics technology and related industries in China, solve technical difficulties faced by enterprises, expand the scale of enterprise promotion, and help enterprises attract more outstanding talents.

In the era of knowledge and information, documents are an important source of information for many cognitive processes such as knowledge database creation, optical character recognition, and document retrieval. As a special and complex entity object, tables convey important information in a very simple way and are widely used in fields such as finance, administration, and archival documents.Using computers to recognize and restore the structure of tables can assist people in processing form information more efficiently and therefore has significant application value.To comprehensively evaluate the performance of the table structure recognition system,this competition has collected table images (including wired and wireless tables) from different scenarios, such as electronic documents, images and so on. Among them, some images have complex backgrounds, image deformations, etc., making this task extremely challenging.

Fig. 1 data used in the competition

Two teams from the University of the Chinese Academy of Sciences and Hikvision (Davar Lab) won the championship and the third place respectively. (Image and text by Lei Hu, initial reviewed by Weiying Yan, reexamined by Shushu Zeng, final reviewd by Jian Zhang)