One Training for Multiple Deployments: Polar-based Adaptive BEV Perception for Autonomous Driving
2023
Source PublicationPROCEEDINGS - IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION
ISSN1050-4729
Volume2023-May
Pages5602-5609
Status已发表
DOI10.1109/ICRA48891.2023.10161552
AbstractCurrent on-board chips usually have different computing power, which means multiple training processes are needed for adapting the same learning-based algorithm to different chips, costing huge computing resources. The situation becomes even worse for 3D perception methods with large models. Previous vision-centric 3D perception approaches are trained with regular grid-represented feature maps of fixed resolutions, which is not applicable to adapt to other grid scales, limiting wider deployment. In this paper, we leverage the Polar representation when constructing the BEV feature map from images in order to achieve the goal of training once for multiple deployments. Specifically, the feature along rays in Polar space can be easily adaptively sampled and projected to the feature in Cartesian space with arbitrary resolutions. To further improve the adaptation capability, we make multi-scale contextual information interact with each other to enhance the feature representation. Experiments on a large-scale autonomous driving dataset show that our method outperforms others as for the good property of one training for multiple deployments. © 2023 IEEE.
KeywordComputing power Large dataset 'current 3D perception Autonomous driving Computing power Computing resource Feature map Large models Learning-based algorithms Regular grids Training process
Conference Name2023 IEEE International Conference on Robotics and Automation, ICRA 2023
Conference PlaceLondon, United kingdom
Conference DateMay 29, 2023 - June 2, 2023
URL查看原文
Indexed ByEI
Language英语
PublisherInstitute of Electrical and Electronics Engineers Inc.
EI Accession Number20233514632065
EI KeywordsAutonomous vehicles
EI Classification Number432 Highway Transportation ; 722.2 Computer Peripheral Equipment ; 722.4 Digital Computers and Systems ; 723 Computer Software, Data Handling and Applications ; 723.2 Data Processing and Image Processing ; 731.6 Robot Applications
Original Document TypeConference article (CA)
Source DataIEEE
Document Type会议论文
Identifierhttps://kms.shanghaitech.edu.cn/handle/2MSLDSTB/325842
Collection信息科学与技术学院
信息科学与技术学院_PI研究组_马月昕
Affiliation
1.School of Information Science and Technology, ShanghaiTech University, Shanghai, China
2.Department of Computer Science and Engineering, Hong Kong University of Science and Technology
3.Department of Information Engineering, The Chinese University of Hong Kong
First Author AffilicationSchool of Information Science and Technology
First Signature AffilicationSchool of Information Science and Technology
Recommended Citation
GB/T 7714
Huitong Yang,Xuyang Bai,Xinge Zhu,et al. One Training for Multiple Deployments: Polar-based Adaptive BEV Perception for Autonomous Driving[C]:Institute of Electrical and Electronics Engineers Inc.,2023:5602-5609.
Files in This Item: Download All
File Name/Size DocType Version Access License
Related Services
Usage statistics
Scholar Google
Similar articles in Scholar Google
[Huitong Yang]'s Articles
[Xuyang Bai]'s Articles
[Xinge Zhu]'s Articles
Baidu academic
Similar articles in Baidu academic
[Huitong Yang]'s Articles
[Xuyang Bai]'s Articles
[Xinge Zhu]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Huitong Yang]'s Articles
[Xuyang Bai]'s Articles
[Xinge Zhu]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 10.1109@ICRA48891.2023.10161552.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.