<abstract xmlns="http://www.w3.org/1999/xhtml">A complex classification task as scene recognition is considered in the present research. Scene recognition tasks are successfully solved by the paradigm of transfer learning from pretrained convolutional neural networks, but a problem is that the eventual size of the network is huge despite a common scene recognition task has up to a few tens of scene categories. Thus, the goal is to ascertain possibility of a size reduction. The modelling recognition task is a small dataset of 4485 grayscale images broken into 15 image categories. The pretrained network is AlexNet dealing with much simpler image categories whose number is 1000, though. This network has two fully connected layers, which can be potentially reduced or deleted. A regular transfer learning network occupies about 202.6 MB performing at up to 92 % accuracy rate for the scene recognition. It is revealed that deleting the layers is not reasonable. The network size is reduced by setting a fewer number of filters in the 17th and 20th layers of the AlexNet-based networks using a dichotomy principle or similar. The best truncated network with 384 and 192 filters in those layers performs at 93.3 % accuracy rate, and its size is 21.63 MB.</abstract>

A complex classification task as scene recognition is considered in the present research. Scene recognition tasks are successfully solved by the paradigm of transfer learning from pretrained convolutional neural networks, but a problem is that the eventual size of the network is huge despite a common scene recognition task has up to a few tens of scene categories. Thus, the goal is to ascertain possibility of a size reduction. The modelling recognition task is a small dataset of 4485 grayscale images broken into 15 image categories. The pretrained network is AlexNet dealing with much simpler image categories whose number is 1000, though. This network has two fully connected layers, which can be potentially reduced or deleted. A regular transfer learning network occupies about 202.6 MB performing at up to 92 % accuracy rate for the scene recognition. It is revealed that deleting the layers is not reasonable. The network size is reduced by setting a fewer number of filters in the 17th and 20th layers of the AlexNet-based networks using a dichotomy principle or similar. The best truncated network with 384 and 192 filters in those layers performs at 93.3 % accuracy rate, and its size is 21.63 MB.

An Efficient Technique for Size Reduction of Convolutional Neural Networks after Transfer Learning for Scene Recognition Tasks

Applied Computer Systems

This work is licensed under the Creative Commons Attribution 4.0 Public License.

A complex classification task as scene recognition is considered in the present research. Scene recognition tasks are successfully solved by the paradigm of...

An Efficient Technique for Size Reduction of Convolutional Neural Networks after Transfer Learning for Scene Recognition Tasks

Published Online: Dec 31, 2018

Page range: 141 - 149

DOI: https://doi.org/10.2478/acss-2018-0018

Keywords
AlexNet, convolutional neural network, pretrained network, scene recognition, size reduction, transfer learning, truncated network

© 2018 Vadim Romanuke, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 Public License.

An Efficient Technique for Size Reduction of Convolutional Neural Networks after Transfer Learning for Scene Recognition Tasks

Published Online: Dec 31, 2018

Page range: 141 - 149

DOI: https://doi.org/10.2478/acss-2018-0018

KeywordsAlexNet, convolutional neural network, pretrained network, scene recognition, size reduction, transfer learning, truncated network

© 2018 Vadim Romanuke, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 Public License.

Keywords
AlexNet, convolutional neural network, pretrained network, scene recognition, size reduction, transfer learning, truncated network