An Efficient Multi-view 3D Object Recognition Mechanism for Distributed Edge Devices

被引:0
|
作者
Yang, Li [1 ]
Hu, Nan [1 ]
Gao, Fei [1 ]
Shen, Gang [1 ]
机构
[1] Nokia Shanghai Bell Co Ltd, Bell Labs, Shanghai 201206, Peoples R China
来源
2022 INTERNATIONAL CONFERENCE ON ADVANCED ROBOTICS AND MECHATRONICS (ICARM 2022) | 2022年
关键词
D O I
10.1109/ICARM54641.2022.9959605
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
This paper proposes an efficient multi-view 3D object recognition mechanism for distributed resource-constrained edge devices. Multi-view based vision processing has various applications on resource-constrained edge devices, like collaborative drone swarms, distributed smart cameras etc. However, huge data transmission from the different edge devices leads to this solution suffering from large bandwidth consumption, long latency, and energy consumption. The efficient multi-view 3D object recognition mechanism includes not only a feature map enhanced scheme, which exploits semantic information to bolster the useful attributes and suppress those less important feature maps, but also the feature maps selection schemes to decrease the amount of data transmission. The proposed fuses and transmits the enhanced feature map, achieves the almost same recognition accuracy with 9-12 times data reduction that can save power consumption and transmission bandwidth compared to the traditional multi-view 3D object recognition mechanism.
引用
收藏
页码:250 / 254
页数:5
相关论文
共 50 条
  • [1] Learning Relationships for Multi-View 3D Object Recognition
    Yang, Ze
    Wang, Liwei
    2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, : 7504 - 7513
  • [2] 3D LayoutCRF for multi-view object class recognition and segmentation
    Hoiem, Derek
    Rother, Carsten
    Winn, John
    2007 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOLS 1-8, 2007, : 580 - +
  • [3] Learning Disentangled Representation for Multi-View 3D Object Recognition
    Huang, Jingjia
    Yan, Wei
    Li, Ge
    Li, Thomas
    Liu, Shan
    IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2022, 32 (02) : 646 - 659
  • [4] Multi-view convolutional vision transformer for 3D object recognition
    Li, Jie
    Liu, Zhao
    Li, Li
    Lin, Junqin
    Yao, Jian
    Tu, Jingmin
    JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION, 2023, 95
  • [5] Multi-view ensemble manifold regularization for 3D object recognition
    Hong, Chaoqun
    Yu, Jun
    You, Jane
    Chen, Xuhui
    Tao, Dapeng
    INFORMATION SCIENCES, 2015, 320 : 395 - 405
  • [6] MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition
    Luequan Wang
    Hongbin Xu
    Wenxiong Kang
    Machine Intelligence Research, 2023, 20 : 872 - 883
  • [7] MVContrast: Unsupervised Pretraining for Multi-view 3D Object Recognition
    Wang, Luequan
    Xu, Hongbin
    Kang, Wenxiong
    MACHINE INTELLIGENCE RESEARCH, 2023, 20 (06) : 872 - 883
  • [8] Multi-view dual attention network for 3D object recognition
    Wenju Wang
    Yu Cai
    Tao Wang
    Neural Computing and Applications, 2022, 34 : 3201 - 3212
  • [9] Deep models for multi-view 3D object recognition: a review
    Alzahrani, Mona
    Usman, Muhammad
    Jarraya, Salma Kammoun
    Anwar, Saeed
    Helmy, Tarek
    ARTIFICIAL INTELLIGENCE REVIEW, 2024, 57 (12)
  • [10] Multi-view dual attention network for 3D object recognition
    Wang, Wenju
    Cai, Yu
    Wang, Tao
    NEURAL COMPUTING & APPLICATIONS, 2022, 34 (04): : 3201 - 3212