Transfer Learning based CNN Model for Personal Authentication using Finger Vein Biometric

Transfer Learning based CNN Model for Personal Authentication using Finger Vein Biometric
Author(s):
SAPNA SHARMA, Dr. Shikha Lohchab
Keywords:
Accuracy , Alex net, Convolutional Neural Network, Google net, Transfer Learning, Squeeze net
Abstract
In this paper, the performance of CNN (convolutional neural networks) such as Alex net, Squeeze net and Google Net (Inception) are analyzed for the finger vein based personal authentication with respect to control, access to the confidential data. The finger vein images from SDUMLA HMT database are used for this research work. Using a wiener filter the noises are removed from the finger vein images. The noise free images are provided for training to Alex net, Squeeze net and Google net network to recognize persons for finger vein authentication. The finger vein authentication using the three pre-trained networks includes loading finger vein images dataset, loading pretrain network, training network through transfer learning, image classification and image validation. The experiment exhibits the outstanding performance of Google net over the Alex net and Squeeze net on several parameters including computation time, the initial learning, accuracy, number of layers and dropout.
Article Details
Unique Paper ID: 153764 Publication Volume & Issue: Volume 8, Issue 8 Page(s): 550 - 555
Article Preview & Download

Share This Article

NCSEM 2024

National Conference on Sustainable Engineering and Management - 2024

Last Date: 15th March 2024

IJIRT.org enables door in research by providing high quality research articles in open access market.

Send us any query related to your research on editor@ijirt.org