中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 自动化研究所
  3. 中国科学院自动化研究所
  4. 多模态人工智能系统全国重点实验室
  5. 先进控制与自动化团队
SASAN: Shape-Adaptive Set Abstraction Network for Point-Voxel 3D Object Detection

文献类型：期刊论文

作者Zhang, Hui1,2; Luo, Guiyang3; Wang, Xiao4; Li, Yidong1,2; Ding, Weiping5; Wang, Fei-Yue6
刊名IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
出版日期2023-12-22
页码15
关键词Feature extraction Three-dimensional displays Detectors Object detection Proposals Point cloud compression Shape 3D object detection autonomous driving point-voxel detectors
ISSN号2162-237X
DOI10.1109/TNNLS.2023.3339889
通讯作者Wang, Xiao(xiao.wang@ahu.edu.cn) ; Li, Yidong(ydli@bjtu.edu.cn)
英文摘要Point-voxel 3D object detectors have achieved impressive performance in complex traffic scenes. However, they utilize the 3D sparse convolution (spconv) layers with fixed receptive fields, such as voxel-based detectors, and inherit the fixed sphere radius from point-based methods for generating the features of keypoints, which make them weak in adaptively modeling various geometrical deformations and sizes of real objects. To tackle this issue, we propose a shape-adaptive set abstraction network (SASAN) for point-voxel 3D object detection. First, the proposal and offset generation module is adopted to learn the coordinates and confidences of 3D proposals and shape-adaptive offsets of the certain number of offset points for each voxel. Meanwhile, an extra offset supervision task is employed to guide the learning of shifting values of offset points, aiming at motivating the predicted offsets to preferably adapt to the various shapes of objects. Then, the shape-adaptive set abstraction module is proposed to extract multiscale keypoints features by grouping the neighboring offset points' features, as well as features learned from adjacent raw points and the 2-D bird-view map. Finally, the region of interest (RoI)-grid proposal refinement module is used to aggregate the keypoints features for further proposal refinement and confidence prediction. Extensive experiments on the competitive KITTI 3D detection benchmark demonstrate that the proposed SASAN gains superior performance as compared with state-of-the-art methods.
WOS关键词CNN
资助项目National Natural Science Foundation of China
WOS研究方向Computer Science ; Engineering
语种英语
WOS记录号WOS:001165506700001
出版者IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
资助机构National Natural Science Foundation of China
源URL[http://ir.ia.ac.cn/handle/173211/57783]  
专题自动化研究所_复杂系统管理与控制国家重点实验室_先进控制与自动化团队
通讯作者Wang, Xiao; Li, Yidong
作者单位1.Beijing Jiaotong Univ, Sch Comp & Informat Technol, Beijing 100044, Peoples R China
2.Beijing Jiaotong Univ, Key Lab Big Data & Artificial Intelligence Transp, Minist Educ, Beijing 100044, Peoples R China
3.Beijing Univ Posts & Telecommun, State Key Lab Networking & Switching Technol, Beijing 100876, Peoples R China
4.Anhui Univ, Engn Res Ctr Autonomous Unmanned Syst Technol, Minist Educ, Hefei 230031, Peoples R China
5.Nantong Univ, Sch Informat Sci & Technol, Nantong 226019, Peoples R China
6.Chinese Acad Sci, Inst Automat, State Key Lab Management & Control Complex Syst, Beijing 100190, Peoples R China
推荐引用方式
GB/T 7714		 Zhang, Hui,Luo, Guiyang,Wang, Xiao,et al. SASAN: Shape-Adaptive Set Abstraction Network for Point-Voxel 3D Object Detection[J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,2023:15.
APA Zhang, Hui,Luo, Guiyang,Wang, Xiao,Li, Yidong,Ding, Weiping,&Wang, Fei-Yue.(2023).SASAN: Shape-Adaptive Set Abstraction Network for Point-Voxel 3D Object Detection.IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS,15.
MLA Zhang, Hui,et al."SASAN: Shape-Adaptive Set Abstraction Network for Point-Voxel 3D Object Detection".IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2023):15.

入库方式： OAI收割

来源：自动化研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。