Classifying and Synthesizing 3D Shapes of Objects using Deep Neural Networks Public Deposited

http://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/b5644x179

Descriptions

Attribute NameValues
Creator
Abstract or Summary
  • Reasoning about 3D shape of objects is important for successful computer visionapplications in robotics, 3D rendering and modeling. In this thesis, we address twoproblems { First, given an image, we generate 3D shape of the foreground object thatappears in the image. Second, we predict the class label of the input 3D object shape.Recent work uses convolutional neural networks (CNNs) for these problems. However,their training is difficult, since it requires a large amount of training data. Also, in 3Dshape generation, existing approaches can not generate realistic 3D shapes and havelimited variations in generated 3D shapes.This thesis addresses these issues. We present two novel approaches, one for eachproblem. First, for 3D classification, we formulate CNN learning as a beam searchaimed at identifying an optimal CNN architecture as well as estimating parametersof such an optimal CNN. This model pursuit approach is evaluated on 3D ShapeNetdataset. Second, we introduce a 3D-VAE-GAN (3D Variational AutoEncoder - Generative Adversarial Network) model to synthesize high-quality 3D objects from 2D imagesin ObjectNet3D dataset. Our experimental evaluation shows that our new CNN learningachieves the state-of-the-art results with better modeling efficiency, i.e., with fewer parameters which are much easier to train. Also, in 3D shape synthesis, we achieve largervariability in shapes.
License
Resource Type
Date Available
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Non-Academic Affiliation
Keyword
Rights Statement
Publisher
Peer Reviewed
Language
Replaces
Additional Information
  • description.provenance : Approved for entry into archive by Steven Van Tuyl(steve.vantuyl@oregonstate.edu) on 2017-06-27T21:23:57Z (GMT) No. of bitstreams: 2license_rdf: 1232 bytes, checksum: bb87e2fb4674c76d0d2e9ed07fbb9c86 (MD5)XuXu2017..pdf: 3692787 bytes, checksum: 4406f9bd36013b1a77d523c0d2236293 (MD5)
  • description.provenance : Approved for entry into archive by Julie Kurtz(julie.kurtz@oregonstate.edu) on 2017-06-23T16:53:16Z (GMT) No. of bitstreams: 2license_rdf: 1232 bytes, checksum: bb87e2fb4674c76d0d2e9ed07fbb9c86 (MD5)XuXu2017..pdf: 3692787 bytes, checksum: 4406f9bd36013b1a77d523c0d2236293 (MD5)
  • description.provenance : Made available in DSpace on 2017-06-27T21:23:57Z (GMT). No. of bitstreams: 2license_rdf: 1232 bytes, checksum: bb87e2fb4674c76d0d2e9ed07fbb9c86 (MD5)XuXu2017..pdf: 3692787 bytes, checksum: 4406f9bd36013b1a77d523c0d2236293 (MD5) Previous issue date: 2017-06-13
  • description.provenance : Submitted by Xu Xu (xuxu@oregonstate.edu) on 2017-06-14T18:01:07ZNo. of bitstreams: 2license_rdf: 1232 bytes, checksum: bb87e2fb4674c76d0d2e9ed07fbb9c86 (MD5)XuXu2017..pdf: 3692787 bytes, checksum: 4406f9bd36013b1a77d523c0d2236293 (MD5)

Relationships

Parents:

This work has no parents.

Last modified

Downloadable Content

Download PDF

Items