International Journal Of Imaging

Abstract

Attendance capture system using face recognition

Author: P.Sri Vaishnavi, N. Lohith, M. Ujwala Qwiny, P. Mahesh Kumar, Ms. Deepika Rathod,

Abstract:

This study aims to improve upon the Visual Geometry Group Algorithm by developing a Face Net Algorithm system that can identify masked faces, as mask usage is now essential due to the COVID-19 epidemic. Materials and Methods: In total of 112 samples, Each group has 56 samples and the number of iterations are 10 for each group. The G power for calculating statistical tests is set at 80%. A total of 2800 images which makes 2240 training images and 560 tested images—make up the research dataset, which is obtained from Kaggle.com combined with self-obtained images. Results: The accuracy results for the Innovative Face Net algorithm is (87.7%), and the Visual Geometry Group (VGG-16) technique is (90.2%). For both methods that were taken into consideration for masked face identification, a 0.001 (p< 0.05) significance value was discovered. Lastly, as compared to the Face Net method, the visual geometry group's (VGG-16) algorithm significantly improves accuracy.