Unsupervised object region proposals for RGB-D indoor scenes

详细信息    查看全文

• 作者：Zhuo Deng ; ^a ; ^{zhuo.deng@temple.edu} ; Sinisa Todorovic^b ; Longin Jan Latecki^a
• 关键词：Object segmentation ; RGB-D ; Sensor fusion
• 刊名：Computer Vision and Image Understanding
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：154
• 期：Complete
• 页码：127-136
• 全文大小：4029 K
• 卷排序：154

文摘

A novel scene structure guided framework for object region proposals is proposed. The RGB-Depth object proposal framework is completely unsupervised. A novel 3D plane segmentation algorithm for predominant plane structures is proposed. The number of proposals is much less than the state-of-the-arts.