method: TILT + Excluding OCR Errors + Fixing total entity2021-02-16

Authors: Research Team


Description: TILT neural network architecture which simultaneously learns layout information, visual features, and textual semantics. Contrary to previous approaches, we rely on a encoder-decoder architecture. We submitted results from a single model. Following the same evaluation procedure as others we removed OCR mismatches and fixed discrepancies of total amount randomly prefixed by "RM".