FormNetV2：面向表单文件信息提取的多模态图形对比学习（英文版）

英文标题：FormNetV2: Multimodal Graph Contrastive Learning for Form Document Information Extraction中文摘要：FormNetV2 引入了一种集中的多模态图形对比学习策略，将自监督预训练统一为一个损失，通过提取与图形边缘相连的一对令牌之间的边界框内的图像特征，捕捉更有针对性的视觉线索，从而在 FUNSD、CORD、SROIE 和 Payment 基准测试上建立新的最先进性能。英文摘要：The recent advent of self-supervised pre-training techniques has led to asurge in the use of multimodal learning in form document understanding.However, existing approaches that extend the mask language modeling to othermodalities require careful multi-task tuni