Login | Register

SdcNet: A Computation-Efficient CNN for Object Recognition


SdcNet: A Computation-Efficient CNN for Object Recognition

Ma, Yunlong and Chunyan, Wang (2018) SdcNet: A Computation-Efficient CNN for Object Recognition. In: DSP 2018.

Text (Final pre-publication version) (application/pdf)
1084_Paper.pdf - Accepted Version
Available under License Spectrum Terms of Access.


Extracting features from a huge amount of data for object recognition is a challenging task. Convolution neural network can be used to meet the challenge, but it often requires a large amount of computation resources. In this paper, a computation-efficient convolutional module, named SdcBlock, is proposed and based on it, the convolution network SdcNet is introduced for object recognition tasks. In the proposed module, optimized successive depthwise convolutions, supported by appropriate data management, are applied in order to generate vectors containing higher density and more varieties of feature information. The hyperparameters can be easily adjusted to suit varieties of tasks under different computation restrictions without significantly jeopardizing the performance. The experiments have shown that SdcNet achieved an error rate of 5.60% in CIFAR-10 with only 55M Flops and also reduced further the error rate to 5.24% using a moderate volume of 103M Flops. The expected computation efficiency of the SdcNet has been confirmed.

Divisions:Concordia University > Gina Cody School of Engineering and Computer Science > Electrical and Computer Engineering
Item Type:Conference or Workshop Item (Paper)
Authors:Ma, Yunlong and Chunyan, Wang
Date:November 2018
Digital Object Identifier (DOI):10.1109/ICDSP.2018.8631567
Keywords:Convolution Neural Network, Object Recognition, Feature Extraction, Successive Depthwise Convolutions, Data Flow Control
ID Code:985308
Deposited By: CHUNYAN WANG
Deposited On:25 Apr 2019 19:15
Last Modified:25 Apr 2019 19:15
All items in Spectrum are protected by copyright, with all rights reserved. The use of items is governed by Spectrum's terms of access.

Repository Staff Only: item control page

Downloads per month over past year

Research related to the current document (at the CORE website)
- Research related to the current document (at the CORE website)
Back to top Back to top