MIVC: Multiple Instance Visual Component for Visual-Language Models

Wenyi Wu, Qi Li, Wenliang Zhong +1

Vision-language models have been widely explored across a wide range of tasks and achieve satisfactory performance. However, it's under-explored how to consolidate entity understan…