ICPR 2024 Competition on Multi-line Mathematical Expressions Recognition

Tianrui Zong, Lin Yang, Wei Luo, Wenhao Yu, Borui Cai, He Zhang, Fenghong Liu, Qinqin Yan, Liangcai Gao

Abstract: Nowadays, mathematical expressions play a significant role in conveying a wide range of information, and accurately recognizing multi-line mathematical expressions is essential for numerous applications. This task is notably challenging due to the intricate spatial relationships between symbols, which complicates the recognition process. Despite its significance, the recognition of multi-line mathematical expressions has received relatively limited attention in both academic literature and competitive evaluations. To advance research and development in this domain, we organized the ICPR 2024 Competition on Multi-line Mathematical Expression Recognition. In this competition, we introduced the AMME-20K dataset, which includes 60% multi-line expressions and more than 4,600 expressions with three or more lines which is over three times the number in the recently published dataset for multi-line expressions recognition task. This competition was held from April 26, 2024 to July 21, 2024 with 9 valid submissions. This report provides a detailed description of the proposed dataset, the task, the evaluation metrics, and the recognition results, highlighting key advancements and insights gained from the competition.

External IDs:dblp:conf/icpr/ZongYLYCZLYG24