Analysis of the efficiency of recognition models in streaming video

This article analyzes VGG, MobileNet, and ResNet architectures for medical face mask recognition in streaming video. Deep convolutional neural network is presented by Python libraries: Tensorflow, Keras, and Haar wavelets based on the models of three architectures and their comparative analysis is p...

Full description

Saved in:
Bibliographic Details
Main Authors: V. M. Goryaev, G. A. Mankaeva, D. B. Bembitov, E. V. Sumyanova, R. A. Bisengaliev
Format: Article
Language:Russian
Published: North-Caucasus Federal University 2025-01-01
Series:Современная наука и инновации
Subjects:
Online Access:https://msi.elpub.ru/jour/article/view/1675
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This article analyzes VGG, MobileNet, and ResNet architectures for medical face mask recognition in streaming video. Deep convolutional neural network is presented by Python libraries: Tensorflow, Keras, and Haar wavelets based on the models of three architectures and their comparative analysis is performed. As a result of the study it was found that the VGGNet method has many times more parameters and is slow to work, while the fastest is Resnet, its accuracy is 97.1% for 30 epochs, and MobileNet - 98.9% for 100 epochs. But according to the results of MobileNet training, it was noticeable that its accuracy increases, and we can conclude that the accuracy will definitely improve if we train on more epochs and photo images, and its compact size allows to use it on devices with modest characteristics.
ISSN:2307-910X