1 paper · 1 filter
Ziyao He, Yingjie Liu, ZhangYangRui +3
Accurate 3D scene description is fundamental to robotic navigation and augmented reality, yet current dense captioning methods face significant limitations in processing sparse poi…