IDEA-Research/Rexverse-2M

ChatRex是一个多模态大型语言模型，它通过结合细粒度对象感知和强大的语言理解能力，能够应对各种需要细粒度感知的场景，如对象检测、grounded对话、图像标注和区域理解。它使用Rexverse-2M数据集，该数据集包含丰富的图像-区域-文本注释。

ChatRex is a Multimodal Large Language Model that integrates fine-grained object perception and robust language understanding to address scenarios such as object detection, grounded conversation, image captioning, and region understanding. It utilizes the Rexverse-2M dataset, which includes diverse image-region-text annotations.