method: Multimodal Transformer for Information Extraction2021-06-02

Authors: Xi Feng, Xiaojie Ming, Meng Guo

Affiliation: ABC Technology


Description: We propose a multimodal transformer architecture for information extraction. The architecture simultaneously fuses coordinate,visual and textual information. Some simple post-processing is applied to SROIE dataset (RM processing in Total). OCR mismatch errors are excluded.