Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

ON THE FEASIBLITY OF SIMULATED DATA FOR 3D OBJECT CLASSIFICATION AND TRANSFER LEARNING

Broderick, Joseph Alois

Abstract Details

, Master of Sciences, Case Western Reserve University, EECS - Computer and Information Sciences.
This thesis describes an approach to enhancing training data for machine learning using simulated perception. Specifically, a deep neural network is trained to recognize 3D objects using a depth camera as input. A challenge is obtaining a sufficiently large training set of labeled images, which can be a laborious manual process. In the present work, it is proposed to enhance a training set with virtual images acquired automatically through use of CAD models of objects viewed by a simulated depth camera. It is shown that a training set augmented with such virtual data can reduce effort in acquiring data while improving network performance.
Wyatt S. Newman (Committee Chair)
M. Cenk Çavuşoğlu (Committee Member)
Gregory S. Lee (Committee Member)

Recommended Citations

Citations

  • Broderick, J. A. (n.d.). ON THE FEASIBLITY OF SIMULATED DATA FOR 3D OBJECT CLASSIFICATION AND TRANSFER LEARNING [Master's thesis, Case Western Reserve University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=case1548325971030332

    APA Style (7th edition)

  • Broderick, Joseph . ON THE FEASIBLITY OF SIMULATED DATA FOR 3D OBJECT CLASSIFICATION AND TRANSFER LEARNING. Case Western Reserve University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=case1548325971030332.

    MLA Style (8th edition)

  • Broderick, Joseph . "ON THE FEASIBLITY OF SIMULATED DATA FOR 3D OBJECT CLASSIFICATION AND TRANSFER LEARNING." Master's thesis, Case Western Reserve University. Accessed APRIL 23, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=case1548325971030332

    Chicago Manual of Style (17th edition)