Hi, @ajjimeno @bdwyer2 @zhxgj @kmh4321 @edwardleardi
Thank you so much for creating such a valuable dataset for the community! I have also studied your great work entitled "Image-based table recognition: data, model, and evaluation". I find that the proposed TEDS metric can optionally take cell content into consideration, i.e., evaluating either table structure recognition or table recognition. However, the results in the paper are only for table recognition task. Could you provide the corresponding "structure_only" TEDS scores in the following table? With these "structure_only" TEDS scores, we can have a more direct and fairer comparison on table structure recognition task, as the OCR engines used by different approaches are quite different. Thank you!
Hi, @ajjimeno @bdwyer2 @zhxgj @kmh4321 @edwardleardi Thank you so much for creating such a valuable dataset for the community! I have also studied your great work entitled "Image-based table recognition: data, model, and evaluation". I find that the proposed TEDS metric can optionally take cell content into consideration, i.e., evaluating either table structure recognition or table recognition. However, the results in the paper are only for table recognition task. Could you provide the corresponding "structure_only" TEDS scores in the following table? With these "structure_only" TEDS scores, we can have a more direct and fairer comparison on table structure recognition task, as the OCR engines used by different approaches are quite different. Thank you!