Learning Inner-Group Relations on Point Clouds
semanticscholar(2021)
摘要
Shown in Fig. 1, we show more examples of attention maps with different scales of grouping. The edge points are more likely to be important in a relatively simple group (i.e., the desk), while for a complex surface, the important points can be anywhere (i.e., the toilet). This observation is reasonable in the real world. To distinguish a shape, we first focus on its outline. But we will consider its internal structure on a complex object.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要