WTW-Dataset
WTW-Dataset copied to clipboard
Performance problem when evaluating Adj-Relation
Sorry for asking problem irrelevant to the dataset itself. But when I'm evaluating Adj-Relation of my results in this dataset, the evaluating code from ICDAR-2019 faced serious performance problem. I checked the evaluation code and found its time complexity too high, when evaluating complex tables like grid tables, it takes too much time and it's unacceptable. I guess you have met the same problem, and ask for help whether you gotta idea to solve this problem. THANK YOU!
Hi, if fact we didn't find the evaluating code, so we follow the rules of ICDAR-2013 and rewrite a evaluating code for Adj-Relation.
May I ask you to consider opensource the adj evaluating code in the future?