الفهرس 
Fundamental ConceptsOnly 14 pages are availabe for public view
 |  | from | 130 |  |  |
| | | from | 130 | | |
Abstract
The purpose of the thesis is to design a monitoring system based on cloud computing that deals with images of people’s faces (in both types: thermal and RGB). These images were taken by special monitoring cameras to citizens while accessing certain public institutions. After uploading the dataset of images to the Cloud, the system processes each image to identify the person’s gender. Then, it extracts a set of features for these people, such as the color of skin and eye color, by using different types of CNN (Resnet 18, 50,101). These features are essential in recognizing faces aiming at simplifying the identification of the public for preventing suspicious persons.
The system targets people wearing masks due to the presence of the Covid 19 virus. It dealt with images of different accuracy. It focuses on excluding important details from the exposed part since a large amount of information is lost due to covering part of the face. Two versions of the system have been coded, the first works offline using Mtalb2020, and the second was online using Python.
The system consists of four phases. The first phase creates a database from photos of each person who is prohibited (infectious patient or intruder). The second phase detects and examines passersby’s eye-forehead region in order to identify them. When individuals from the uploaded database appear in any monitored section, the third phase recognizes them as banned. The essential alert procedures for prevention are implemented in phase four.
The model achieves a very good accuracy percentage in contrast to loss. Tests reveal high ratios for variant performance measures. We believe this work will be a demand, for safety precautions, especially in vital enterprises. Because recognizing the identity of masked persons is essential, as the problem difficulty increased since the start of the Covid-19 pandemic.
This thesis consists of six chapters.
Chapter 1:
Background information on the fundamental subjects covered in this thesis is provided in this chapter. First, it introduces the concept of Cloud computing and provides an introduction to its architecture, models, traits, and applications. Second, our research topic which is monitoring systems is introduced by examining their architectures, key characteristics, and detection of visual surveillance systems. Third, the Neural Network, the core component of the deep learning technique used in this thesis, is discussed, focusing on convolutional neural networks. Fourth, the challenges of face detection with its most widely used application systems are covered.
Chapter 2:
This chapter surveys the majority of the thesis-relevant research. The introduced feature extraction strategies are summarized. Machine learning gender classifiers for masked individuals are highlighted. The methods for classifying facial thermal pictures are covered. Some monitoring methods based on cloud infrastructure are reviewed.
Chapter 3:
This chapter is organized as follows. It begins by highlighting the thesis motivation. After that, it presents the suggested model, labeled by IRT-ResNet, for classifying persons depending on their gender. This process is based on a deep ResNet convolutional neural network model using infrared thermal images. Finally, it provides an overview of the experiments that were done, to gauge effectiveness, as well as the results of the comparisons.
Chapter 4:
This chapter explores our model built for analyzing masked face RGB images using CNN to identify gender using cloud computing and to detect eyes and skin color. It begins with explaining the motivation behind the proposal. Then, it moves on to introducing the detection algorithm that has been altered to accommodate our suggestion. Following, it describes the convolutional neural network-based recognition method that is suggested for determining an individual’s gender, skin tone, and eye color from their possession of an image. Also, it makes a suggestion for a new database for masked faces and describes how to recognize faces and extract the necessary details. At the end, it studies the operated tests and their outcomes using the Cloud.
Chapter 5:
This chapter explains the main objective of the thesis, which is to build the proposed monitoring system based on cloud computing to identify banned persons, whether infected or intruders. This is in order to prevent them from entering government institutions or public enterprises. And their data is displayed on the screen to prevent them, as well as to send an alert by e-mail. The proposed work methodology is presented, then the basic system design stages. At the end, the results obtained from the experiments that have been conducted are figured and discussed.
Chapter 6:
This chapter contains the conclusion of the thesis and future works.