MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition

doi:10.1007/s11633-023-1430-z

CASIA OpenIR > 学术期刊 > Machine Intelligence Research

	MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition
	Luequan Wang; Hongbin Xu; Wenxiong Kang
发表期刊	Machine Intelligence Research
ISSN	2731-538X
	2023
卷号	20 期号:6 页码:872-883
摘要	3D shape recognition has drawn much attention in recent years. The view-based approach performs best of all. However, the current multi-view methods are almost all fully supervised, and the pretraining models are almost all based on ImageNet. Although the pretraining results of ImageNet are quite impressive, there is still a significant discrepancy between multi-view datasets and ImageNet. Multi-view datasets naturally retain rich 3D information. In addition, large-scale datasets such as ImageNet require considerable cleaning and annotation work, so it is difficult to regenerate a second dataset. In contrast, unsupervised learning methods can learn general feature representations without any extra annotation. To this end, we propose a three-stage unsupervised joint pretraining model. Specifically, we decouple the final representations into three fine-grained representations. Data augmentation is utilized to obtain pixel level representations within each view. And we boost the spatial invariant features from the view level. Finally, we exploit global information at the shape level through a novel extract-and-swap module. Experimental results demonstrate that the proposed method gains significantly in 3D object classification and retrieval tasks, and shows generalization to cross-dataset tasks.
关键词	Multi view, unsupervised pretraining, contrastive learning, 3D vision, shape recognition
DOI	10.1007/s11633-023-1430-z
引用统计	被引频次：3[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/56015
专题	学术期刊_Machine Intelligence Research
作者单位	School of Automation Science and Engineering, South China University of Technology, Guangzhou 510641, China
推荐引用方式 GB/T 7714	Luequan Wang,Hongbin Xu,Wenxiong Kang. MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition[J]. Machine Intelligence Research,2023,20(6):872-883.
APA	Luequan Wang,Hongbin Xu,&Wenxiong Kang.(2023).MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition.Machine Intelligence Research,20(6),872-883.
MLA	Luequan Wang,et al."MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition".Machine Intelligence Research 20.6(2023):872-883.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
MIR-2022-11-334.pdf（1954KB）	期刊论文	出版稿	开放获取	CC BY-NC-SA	浏览下载