Enhancing Visual Document Understanding with Contrastive Learning in Large Visual-Language Models

arXiv:2402.19014v1 Announce Type: new
Abstract: Recently, the advent of Large Visual-Language Models (LVLMs) has received increasing attention across various domains, particularly in the field of visual document understanding (VDU). Different from conventional vision-language tasks, VDU is specifically concerned with text-rich scenarios containing abundant document elements. Nevertheless, the importance of fine-grained features remains largely unexplored within the community of LVLMs, leading to suboptimal performance in text-rich scenarios. In this paper, we abbreviate it as the fine-grained feature collapse issue. With the …

abstract arxiv attention cs.cv document document understanding domains importance language language models tasks text type understanding vision visual