Can Hallucination Correction Improve Video-Language Alignment?

Lingjun Zhao, Mingyang Xie, Paola Cascante-Bonilla +2

Large Vision-Language Models often generate hallucinated content that is not grounded in its visual inputs. While prior work focuses on mitigating hallucinations, we instead explor…