logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

Multiple-view D2NNs array: realizing robust 3D object recognition

delete2021-07-09
delete19
PRE
AI
史珈硕 cover
史珈硕 (Jiashuo Shi)
L
Liang Zhou
T
Taige Liu
C
Chai Hu
K
Kewei Liu
J
Jun Luo
H
Haiwei Wang
C
Changsheng Xie
张新宇 cover
张新宇 (Xinyu Zhang) *
DOI：10.1364/OL.432309delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
As an optical-based classifier of the physical neural network, the independent diffractive deep neural network ((DNN)-N-2) can be utilized to learn the single-view spatial featured mapping between the input lightfields and the truth labels by preprocessing a large number of training samples. However, it is still not enough to approach or even reach a satisfactory classification accuracy on three-dimensional (3D) targets owing to already losing lots of effective light-field information on other view fields. This Letter presents a multiple-view D(2)NNs array (MDA) scheme that provides a significant inference improvement compared with individual (DNN)-N-2 or Res-(DNN)-N-2 by constructing a different complementary mechanism and then merging all base learners of distinct views on an electronic computer. Furthermore, a robust multiple-view D(2)NNs array (r-MDA) framework is demonstrated to resist the redundant spatial features of invalid lightfields due to severe optical disturbances. (C) 2021 Optical Society of America.
Keywords:
LIGHT

Journal

Optics Letters cover
Optics Letters
IF:
3.3
Papers:
4.0W
Citations:
7.6W

Organization

No organization information available