Multi-view Convolutional Neural Networks for 3D Shape Recognition

被引:2512
作者
Su, Hang [1 ]
Maji, Subhransu [1 ]
Kalogerakis, Evangelos [1 ]
Learned-Miller, Erik [1 ]
机构
[1] Univ Massachusetts, Amherst, MA 01003 USA
来源
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV) | 2015年
基金
美国国家科学基金会;
关键词
D O I
10.1109/ICCV.2015.114
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
A longstanding question in computer vision concerns the representation of 3D shapes for recognition: should 3D shapes be represented with descriptors operating on their native 3D formats, such as voxel grid or polygon mesh, or can they be effectively represented with view-based descriptors? We address this question in the context of learning to recognize 3D shapes from a collection of their rendered views on 2D images. We first present a standard CNN architecture trained to recognize the shapes' rendered views independently of each other, and show that a 3D shape can be recognized even from a single view at an accuracy far higher than using state-of-the-art 3D shape descriptors. Recognition rates further increase when multiple views of the shapes are provided. In addition, we present a novel CNN architecture that combines information from multiple views of a 3D shape into a single and compact shape descriptor offering even better recognition performance. The same architecture can be applied to accurately recognize human hand-drawn sketches of shapes. We conclude that a collection of 2D views can be highly informative for 3D shape recognition and is amenable to emerging CNN architectures and their derivatives.
引用
收藏
页码:945 / 953
页数:9
相关论文
共 36 条
[1]  
[Anonymous], P S GEOM PROC
[2]  
[Anonymous], 2013, P 31 INT C MACHINE L
[3]  
[Anonymous], 2014, P BMVC
[4]  
[Anonymous], 2014, P IEEE C COMPUTER VI
[5]  
[Anonymous], DEEPVISION WORKSH
[6]  
[Anonymous], P ECCV
[7]  
[Anonymous], 2015, P CVPR
[8]  
[Anonymous], 2005, P CVPR
[9]  
[Anonymous], P CVPR
[10]  
[Anonymous], P CVPR