1 paper · 1 filter
Jianxin Huang, Jiahang Li, Sergey Vityazev +2
RGB-D scene parsing methods effectively capture both semantic and geometric features of the environment, demonstrating great potential under challenging conditions such as extreme…