TY - GEN
T1 - 3D Capsule Networks for Object Classification from 3D Model Data
AU - Ahmad, Ayesha
AU - Kakillioglu, Burak
AU - Velipasalar, Senem
N1 - Funding Information:
The information, data, or work presented herein was funded in part by National Science Foundation (NSF) under Grant 1739748 and by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award Number DE-AR0000940. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
Publisher Copyright:
© 2018 IEEE.
PY - 2019/2/19
Y1 - 2019/2/19
N2 - Many of the existing object classification methods today rely on convolutional neural networks (CNNs), which are very successful in extracting features from the data. However, CNNs cannot sufficiently address the spatial relationship between features and require large amounts of data for training. In this paper, a new architecture is proposed for 3D object classification, which is an extension of the Capsule Networks (CapsNets) to 3D data. Our proposed 3D CapsNet architecture preserves the orientation and spatial relationship of the extracted features, and thus requires less data to train the network. We compare our approach with a ShapeNet inspired model, and show that our method provides performance improvement especially when training data size gets smaller. We also compare and evaluate several different versions of the 3D Capsnet architecture.
AB - Many of the existing object classification methods today rely on convolutional neural networks (CNNs), which are very successful in extracting features from the data. However, CNNs cannot sufficiently address the spatial relationship between features and require large amounts of data for training. In this paper, a new architecture is proposed for 3D object classification, which is an extension of the Capsule Networks (CapsNets) to 3D data. Our proposed 3D CapsNet architecture preserves the orientation and spatial relationship of the extracted features, and thus requires less data to train the network. We compare our approach with a ShapeNet inspired model, and show that our method provides performance improvement especially when training data size gets smaller. We also compare and evaluate several different versions of the 3D Capsnet architecture.
KW - 3D
KW - Capsule networks
KW - object classification
UR - http://www.scopus.com/inward/record.url?scp=85063006154&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85063006154&partnerID=8YFLogxK
U2 - 10.1109/ACSSC.2018.8645256
DO - 10.1109/ACSSC.2018.8645256
M3 - Conference contribution
AN - SCOPUS:85063006154
T3 - Conference Record - Asilomar Conference on Signals, Systems and Computers
SP - 2225
EP - 2229
BT - Conference Record of the 52nd Asilomar Conference on Signals, Systems and Computers, ACSSC 2018
A2 - Matthews, Michael B.
PB - IEEE Computer Society
T2 - 52nd Asilomar Conference on Signals, Systems and Computers, ACSSC 2018
Y2 - 28 October 2018 through 31 October 2018
ER -