Global Patch Cross-Attention for Point Cloud Analysis

doi:https://doi.org/10.1007/978-3-031-18913-5_8

CASIA OpenIR > 紫东太初大模型研究中心

	Global Patch Cross-Attention for Point Cloud Analysis
	Tao ML(陶满礼)1,2 ; Zhao CY(赵朝阳)2 ; Wang JQ(王金桥)1,2 ; Tang M(唐明)1,2
	2022-10
会议名称	Pattern Recognition and Computer Vision (PRCV 2022)
会议录名称	5th Chinese Conference, PRCV 2022, Shenzhen, China, November 4–7, 2022, Proceedings, Part III
卷号	3
页码	96-111
会议日期	2022.11.4-2022.11.7
会议地点	深圳
会议录编者/会议主办者	Shiqi Yu, Zhaoxiang Zhang, Pong C. Yuen, Junwei Han, Tieniu Tan, Yike Guo, Jianhuang Lai, Jianguo Zhang
出版地	Cham
出版者	Springer
摘要	Despite the great achievement on 3D point cloud analysis with deep learning methods, the insufficiency of contextual semantic description, and misidentification of confusing objects remain tricky problems. To address these challenges, we propose a novel approach, Global Patch Cross-Attention Network (GPCAN), to learn more discriminative point cloud features effectively. Specifically, a global patch construction module is developed to generate global patches which share holistic shape similarity but hold diversity in local structure. Then the local features are extracted from both the original point cloud and these global patches. Further, a transformer-style cross-attention module is designed to model cross-object relations, which are all point-pair attentions between the original point cloud and each global patch, for learning the context-dependent features of each global patch. In this way, our method can integrate the features of original point cloud with both the local features and global contexts in each global patch for enhancing the discriminative power of the model. Extensive experiments on challenging point cloud classification and part segmentation benchmarks verify that our GPCAN achieves the state-of-the-arts on both synthetic and real-world datasets.
关键词	Global patch · Cross-attention · Contextual description · Point cloud analysis
学科门类	工学::控制科学与工程
DOI	https://doi.org/10.1007/978-3-031-18913-5_8
URL	查看原文
收录类别	EI
语种	英语
是否为代表性论文	是
七大方向——子方向分类	三维视觉
国重实验室规划方向分类	视觉信息处理
是否有论文关联数据集需要存交	否
引用统计
文献类型	会议论文
条目标识符	http://ir.ia.ac.cn/handle/173211/56736
专题	紫东太初大模型研究中心
通讯作者	Tao ML(陶满礼)
作者单位	1.中国科学院大学 2.中国科学院自动化研究所
第一作者单位	中国科学院自动化研究所
通讯作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Tao ML,Zhao CY,Wang JQ,et al. Global Patch Cross-Attention for Point Cloud Analysis[C]//Shiqi Yu, Zhaoxiang Zhang, Pong C. Yuen, Junwei Han, Tieniu Tan, Yike Guo, Jianhuang Lai, Jianguo Zhang. Cham:Springer,2022:96-111.

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
GPCAN.pdf（6422KB）	会议论文		开放获取	CC BY-NC-SA	浏览下载