Detection Dataset

该数据集能够生成关于整个场景的对话，提供了场景中每个物体的位置信息。回应形式是以场景中所有物体的边界框列表来制定。该数据集是在现有行为克隆数据的基础上进行规模扩展的，其任务是进行场景理解。

This dataset is capable of generating dialogues about the entire scene, and provides the positional information of each object within the scene. The response format is formulated using a list of bounding boxes of all objects in the scene. This dataset is scaled up based on existing behavioral cloning data, with its core task being scene understanding.