Personal Authentication Application Using Deep Learning Neural Network

Biometric authentication method based on image processing has obtained considerable attention. Images captured have its unique features especially the variance between objects and background. Hence n this paper, an automatic approach for personal recognition using deep learning algorithm namely the Basic Alexnet model that utilized the SVM algorithm for classification is proposed. The proposed algorithm consists of the following: data acquisition, designing the architecture of the Deep Learning Model (DLM), training the DLM, testing the DLM followed by training and testing the DLM in real time. The proposed biometric authentication method can be used in hospitals and for forensic applications. The total images acquired are 900. Training images are 70% of total images (630 images) and the remainder as testing images for training and testing the basic Alexnet model. Results attained showed that the highest accuracy is at 98%.

Paper

Full text

PDF

Personal Authentication Application Using Deep Learning Neural Network

Semantic Scholar · Computer Science · 2020

Abstract

Biometric authentication method based on image processing has obtained considerable attention. Images captured have its unique features especially the variance between objects and background. Hence n this paper, an automatic approach for personal recognition using deep learning algorithm namely the Basic Alexnet model that utilized the SVM algorithm for classification is proposed. The proposed algorithm consists of the following: data acquisition, designing the architecture of the Deep Learning Model (DLM), training the DLM, testing the DLM followed by training and testing the DLM in real time. The proposed biometric authentication method can be used in hospitals and for forensic applications. The total images acquired are 900. Training images are 70% of total images (630 images) and the remainder as testing images for training and testing the basic Alexnet model. Results attained showed that the highest accuracy is at 98%.

Similar papers

© 2026 NYSGPT2525 LLC