SqueezeVGGNet: A Methodology for designing low complexity VGG Architecture for Resource Constraint Edge Applications

Chandrapu, Ramesh Reddy and Pal, Chandrajit and Nimbekar, Anagha Tilak and Acharyya, Amit (2022) SqueezeVGGNet: A Methodology for designing low complexity VGG Architecture for Resource Constraint Edge Applications. In: 20th IEEE International Interregional NEWCAS Conference, NEWCAS 2022, 19 June 2022 through 22 June 2022, Quebec City.

[img] Text
20th_IEEE_International.pdf - Published Version
Restricted to Registered users only

Download (2MB) | Request a copy


Convolutional Neural Networks also known as ConvNets are now extensively used in various machine learning tasks for solving various problems in computer vision, biomedical, defence industry, entertainment etc. These neural networks for most of the applications are focused towards increasing the accuracy. However, besides maintaining the accuracy within a tolerable range, reduction in the network model size can have a lot of advantages from its mobility, easy deployment, remote upgradation and energy efficiency point of view. To attain these advantages, we propose a universal strategy to realize the convolution operation of a n x n filter kernel with fewer parameters, which also reduces the number of channels. We have proposed a compressed VGGNet model based on VGGNet neural network which resulted in 20x lesser parameters compared to its classical counterpart with an improved inference time by 3 times whilst maintaining similar accuracy. A complete hardware design of the compressed VGG architecture has also been implemented. A quantitative and qualitative analysis for various variants of VGGNet and other models reveal the reduction in the number of parameters in the range of 18-20x and the number of network operations contributing to the model complexity has shown a reduction of 2.5x with respect to its vanilla counterpart making it easier to deploy onto FPGAs and edge devices © 2022 IEEE.

[error in script]
IITH Creators:
IITH CreatorsORCiD
Acharyya, Amithttp://orcid.org/0000-0002-5636-0676
Item Type: Conference or Workshop Item (Paper)
Additional Information: The authors like to specially thank the contribution of Mr Srikar YM and Sai Niranjan Reddy for their ideas and constant support in making this project successful. This project is partially funded by Defence Research Development Organization. The CAD tools and reconfigurable platforms are supported by Ministry of Electronics and Information Technology, Govt of India.
Uncontrolled Keywords: CNN (Convolutional Neural Network); compaction; COMPEXP; expansion; squeezeVGGNet
Subjects: Electrical Engineering
Divisions: Department of Electrical Engineering
Depositing User: . LibTrainee 2021
Date Deposited: 08 Oct 2022 07:04
Last Modified: 08 Oct 2022 07:04
URI: http://raiith.iith.ac.in/id/eprint/10854
Publisher URL: http://doi.org/10.1109/NEWCAS52662.2022.9841955
Related URLs:

Actions (login required)

View Item View Item
Statistics for RAIITH ePrint 10854 Statistics for this ePrint Item