Faculty Of Computer Science Graduation Project

Permanent URI for this communityhttp://185.252.233.37:4000/handle/123456789/41

Browse

Now showing 1 - 20 of 130

Reassemble Destroyed Paper
(October University for Modern Sciences and Arts, 2018) Youssef, Heba
This project looks at the challenges of creating the automatic system for reconstruction shredded documents cut vertically (strip) and vertically, horizontally (cross- cut). The un-shredding problem is of interest in the fields of forensics, investigative sciences, archaeology and criminal investigation. All stages of the un-shredding pipeline are analysed, starting from uploaded images of shreds and ending with reconstructed documents. The current bottlenecks in this pipeline are identified and solutions are proposed. At the beginning a formal definition of this problem and references to related work will be given. While there are some approaches published dealing with the reconstruction of destroyed paper in general, there is barely work done in the field of cross-cut shredding then giving an introduction to algorithm of ant-colony. The solutions proposed in this project start with narrowing problem then implement two reconstruction approaches to improve the results. First approach, Character database, Feature matching algorithm, Row Clustering, Estimate error evaluation function and Greedy classifier then get the results. To improve the results that belongs to first approach we implement second approach. Second approach is reconstructing by Ant colony algorithm by estimate different error evaluation function then apply the algorithm.
Playing Music using Contactless Gestures
(October university for modern sciences and arts, 2018) Kassem, Salma Khaled
Throughout history and since the beginning of times, people have created various musical instruments suited for their own traditional music and culture. These various instruments gave birth to many beautiful musical notes that live up until today. Therefore, regardless of one’s cultural background or language, we can safely assume that everyone enjoys a certain style of music. However, since musical instruments originated roughly around 37,000 years ago (where there was absolutely no technology), the only way to play music has always been by using one’s hands. This way of playing an instrument has been passed down all the way until this day. Even though music has been developing since the day it originated up until today, people have created ways that make music sound different. Nowadays, a lot of, if not most, of the music we hear today is either computerized, or altered by a computer. However, people have only focused on altering the way the music sounds, but there was little, if no, attention given to the way music is played. Moreover, music is known to have a lot of positive side effects on the practitioner as well as the listener; but we are more focused on the practitioner in this paper. Nowadays a lot of people are willing to learn how to play musical instruments such as the guitar, the drums, the piano, etc. However, a lot of people are either unable to have access to musical facilities or are unable to purchase a musical instrument for the sole purpose of personal entertainment. Therefore, a lot of people slack on playing musical instruments or just fail at it. Moreover, some musicians try to look for creative new techniques to experiment with music 4 | P a g e electronically in order to improve the quality of their music or simply just to create something new. However, there are very few creative ways to play musical instruments. This project aims to make playing and learning music easy and accessible to everyone everywhere while also being a new one of a kind experience.
Multivariate analysis & Finding the Origin of olive
(October university for modern sciences and arts, 2018) Khalifa, Khalifa Sebak
Spectroscopy and the information obtained from the level of energy, allows researchers to extract analytical information about certain matter and the relation between them Multivariate Regression Analysis is a technique that uses more than one variable and estimates one single regression model along with these variables, there are more than one predictor variable which used to predict other variables in the regression model, it can be used if there are 3 more dependent variables to examined against independent variables or just one variable. We used in order to understand how the variables are influenced by other factors and to what extent they are related to each other, As an application for this great technology in this project detecting olive origin would be great example for how to use these technologies, classification by Principle component analysis (PCA), and partial least square regression (PLSR) are used along with this application.
Silhouette - Based Face Recognation
(October University for Modern Sciences and Arts, 2018) Ahmed, Maryam Ahmed
A silhouette face is representing the face as one solid shaped color, usually, the color is dark and removes every inline detail of the face but the silhouette’s outline edges are matching the original face’s outline. The silhouette of the face is usually the outline of the face profile, and face profile is considered a very important aspect of face identification it also provides an overall structure view of the face that is seen in the non-frontal view. Normally side view images are captured from a distance which unfortunately leads to capturing the image in a low-resolution quality. The main purpose of this project is to introduce a combination of silhouette detection and side-view face detection in order to detect, process, and compare a person from a side-view image. In this thesis we will be moving step by step starting fromthe step of extracting the face’s outer line from the side-view colored face picture. And start deploying methods and algorithms on these outer line points in order to approach the goal of matching the person with one of the known and saved persons we have. This thesis is introducing a module that can be added to face recognition systems in order to raise the recognition accuracy.
Supervised Machine Learning in Cancer Diagnosis
(October University of Modern Sciences and Arts, 2018) Ahmed, Mahmoud Sayed
Mathematics is the study of natural patterns which offers ways and tools to describe these patterns. In other words, mathematics tries to find a set of rules which mimics the behavior of a repeated or regular way in which something happens or done. In this dissertation we are going to research the patterns, in which breast cancer function, focus on Breast Cancer as it’s the most lethal type of cancer in Egypt which takes the lives of 10,000 female each year according to the World Health Organization. i Since Invasive Ductal Carcinoma (IDC) is the most common type of breast cancerii , this dissertation chooses the IDC to be its use-case in studying patterns. We focus in recognizing the IDC mathematically without human-intervention, which can automate the process of diagnosing IDC breast cancer. We used the latest technologies of Deep Neural Networks (DNN) and Convolutional Neural Networks (CNN) as well as classical machine learning methods as K-Nearest-Neighbor (KNN), Support-Vector- Machine (SVM), Random Forest and many others and we eventually obtained an accuracy of 85% using 90K data samples; this data were so complex and this is so close of the current state-of-the-art implementation on this data which is 87%. While doing so, we will watch our own behavior while researching a certain pattern, and tries to conclude a pattern from it, which ultimately can produces a product or an application to help doctors who are researching cancer to use it and try to find patterns mathematically even without them having programming or mathematical background.
Interactive Board Writing On Board Using Pointer Finger
(October university for modern sciences and arts, 2018) Sobhy, Rana Ahmed
Human-Computer Interaction (HCI) between a human and a computer can be used to develop a user-friendly interface. Gestures are one of the ways to interact non-verbally in HCI. Gesture recognition is used to convey information to a system or to control it; therefore it can replace some of the hardware devices used to control computer activities. Recognizing gestures in real-time became more feasible due to the advancements in the fields of pattern recognition and Image Processing. This is a documentation of a project that implements an algorithm that can be used to track and recognize hand gestures for drawing on a board. This algorithm depends on three steps: hand tracking, hand detection, and fingertip detection. For hand tracking the Mean- Shift algorithm is used, which is used to track non-rigid objects based on their histogram. For the hand segmentation, a simple background subtraction and threshold is used and it is also used to detect the fingertip of the pointer finger.
Sky battle : 3D open GL PC game
(October University of Modern Sciences and Arts, 2018) Nour El Din, Ahmed Wael Mohammad
Open Graphics Library (OpenGL) is an application programming interface (API) for rendering 2D and 3D Graphics. This API is usually used for rendering Graphics using Graphics processing unit (GPU) in order to achieve more efficient and optimized rendering. OpenGL is used in several zones including video games, computer-aid design (CAD), virtual reality (VR) scientific visualization, flight simulation. OpenGL uses OpenGL shading language (GLSL), which is responsible for calculating the final color and brightness of each pixel after applying light, shadows, reflections, etc. My aim is to create a high-quality 3D realistic game based on aircraft battle, with as less bugs and crashes as possible. The game is controlled using mouse and keyboard and processes fog, shadows, multiple light sources and some special light effects in order to make the game more realistic. Apart from visuals, the game functions are focused on collision detection, some physics and controls along with camera movement. The game uses lightweight java game library (LWJGL) that enables cross- platform access APIs useful in developing graphics like OpenGL when developing in java.
Voice Controlled Robotic Arm Manipulation
(October University for Modern Sciences and Arts, 2018) Ezz, Youssef Khaled
A robot is an assemblage of a certain material that works or moves in a way that imitates the ways of living creatures and organisms. There are many applications of robots, which mostly use their available sensors to do a certain jobs, or use its actuators to affect the environment in some way or another. The main reason behind creating a robot is to facilitate the life of a human being. In other words to make things done faster and more accurately without any human intervention. This work aims to create a robotic arm manipulation controlled by voice command, which could be integrated into a higher scale model to help into construction or into hazardous environments. This robot consists of servo motors as its joints and metal for its links. The end effector of the arm is a gripper that can hold objects, stack them or bring them down to the ground through voice commands. The movement of the robot is calculated using forward kinematics and inverse kinematic.
Prediction of thermal comfort by using machine learning
(October University for Modern Sciences and Arts, 2019) Elsebaie Ahmed, Youssef
When working in a workplace whether under high or low temperature it reduces work performance. As thermal comfort plays a significant role for indoor environment when it comes to work in work places. The aim of this study is to propose a method by using a machine learning algorithm to predict the thermal comfort. Neural network’s algorithms will be implemented by using the momentum for the dataset required. After that taking the final weights and then switching to android studio to read the data coming from the Arduino-Uno device via Bluetooth and use the resulted final weights for classification and start predicting the thermal comfort. The proposed solution consists of 4 steps: 1) collect data, 2) feature extraction, 3) classification, 4) evaluation. external components were used such as: 1) Arduino-Uno, 2) HC05-bluetooth, 3) LM-35 temperature sensors just to connect them to the application and start receiving the temperature. After the application is developed it will help the user with the current and predicted thermal comfort/discomfort using mobile application. The output accuracy was that thermal “Comfort” is 72.7% and only 28.3% remained for thermal “discomfort” And for later future work, Relative humidity and air velocity sensors will be implemented to get the correct data for each. أثناء العمل في مكان، سواء تحت درجة حرارة عالية أو منخفضة ، فإنه يقلل من أداء العمل. كما تلعب الراحة الحرارية دورًا مهمًا للبيئة الداخلية عندما يتعلق الأمر بالعمل في الأماكن الخاصة. الهدف من هذه الدراسة هو اقتراح طريقة باستخدام خوارزمية التعلم الآلي للتنبؤ بالراحة الحرارية. سيتم تنفيذ خوارزميات الشبكة العصبية باستخدام الزخم لمجموعة البيانات المطلوبة. بعد ذلك أخذ الأوزان النهائية ثم الانتقال إلى android studio لقراءة البيانات الواردة من جهاز Arduino-Uno عبر البلوتوث واستخدام الأوزان النهائية الناتجة للتصنيف والبدء في التنبؤ بالراحة الحرارية. يتكون الحل المقترح من 4 خطوات: 1) جمع البيانات ، 2) استخراج الميزة ، 3) التصنيف ، 4) التقييم. تم استخدام المكونات الخارجية مثل:) Arduino-Uno(1 ، HC05-bluetooth(2 ، 3) أجهزة استشعار درجة الحرارة LM-35 فقط لتوصيلها بالتطبيق والبدء في تلقي درجة الحرارة بعد تطوير التطبيق ، سيساعد المستخدم في توفير الراحة / الانزعاج الحراري الحالي والمتوقع باستخدام تطبيق الهاتف المحمول. كانت دقة الخرج هي أن الراحة الحرارية هي 72.7٪ وبقيت 28.3٪ فقط بسبب الانزعاج الحراري ، وبالنسبة للعمل في المستقبل ، سيتم تطبيق مستشعرات الرطوبة النسبية وسرعة الهواء للحصول على البيانات الصحيحة ككل
Unreal Engine Game with Use of AI
(October University for Modern Sciences and Arts, 2019) Turchenko, Illia
Games have existed and been an intrinsic part of civilization since its' very beginning. People have used games to learn and be entertained. The era of computers has created new endless possibilities for expressing creativity and creating interactive games for many to enjoy. Gaming has always peaked interest of the public, and it is ever growing, especially now with the advancements in mobile technology and its' wide access to the public, which allows to expand the gamer community even further. The field has come a long way since the days of simple 8-bit installments, which is largely attributed to the technological advancements in graphics, audio, AI etc. To meet the ever-increasing complexity of games, various game engines have been developed to assist developers in their task of creating games. The project is a game developed using Unreal Engine, one of the most ambitious and popular engines to date. The game is an action-adventure role-playing game, where you control a humanoid character in an open world. You possess immersive combat system, which may change depending on the circumstances, such as combat mode and power level. You are able to control and interact with vehicles, which change the controls and physics attached, embark on missions, which challenge your skills as a gamer, and more. The world is full of smart enemy AI, which may share components with the player or vehicles, in order to provide a fair experience. The game features a progression system for your character, which allows you to collect experience and level up, growing in power. كانت الألعاب موجودة وكانت جزء لا يتجزأ من الحضارة منذ بدايتها ، لقد إستخدم الناس الألعاب للتعلم والإستمتاع . لقد أوجد عصر أجهزة الكمبيوتر إمكانيات جديدة لا نهاية لها للتعبير عن الإبداع وإنشاء ألعاب تفاعلية يستمتع بها الكثيرون
Property finding mobile application with smart analytics (E property)
(October University for Modern Sciences and Arts, 2019) Ibrahim Abouelfetouh Hegazy, Mohamed
Do you dream of owning property? Perhaps multiple investment properties from which to earn a monthly stream of income? But before you contact your real estate agent , consider what’s really involved. If you want to create an income immediately, you’ll need to rent your property. Though the proposition may sound simple, it is anything but. The information that follows details the downsides to property management. It’s not meant to dissuade you from moving forward, but instead to show you that there are downsides as well as advantages to buying and managing rental property. Don’t let the potential to earn money cloud your vision when considering whether or not you’re cut out for it.
Flight Simulator with autopilot capabilities
(2019) Ali Fayed, Youssef
Flight simulators are thought as a game but in reality this is so far away from the truth, flight simulators are really important as it is used in so many levels of the flying industries, it is important in the flight safety, aircraft manufacturers and flying agencies all around the world spend tremendous amounts of money to develop flight simulators that can imitate real life as close as possible, for example in 2012 Microsoft spent 4.2 million to develop and upgrade its flight simulator game to a real flight simulator to train new pilots on the basics of aerodynamics of planes. Multiple giant aircraft makers are trying to develop an Automated flight system (AFS) that can truly fly the plane from one end to another as the current autopilots can only do limited amount of tasks and can only fly the plane in a limited conditions due to the limited computational power and technologies available, but currently machine learning are so popular and powerful that we can finally try to achieve that dream by training a model to fly a plane without the need for any human interaction. Summary: يُنظر إلى أجهزة محاكاة الطيران على أنها لعبة ولكن في الواقع هذا ابعد ما يكون عن الحقيقه ، فإن أجهزة محاكاة الطيران مهمة جدًا لأنها تستخدم في العديد من مستويات صناعة الطيران ، فهي مهمة في سلامة الطيران ومصنعي الطائرات ووكالات الطيران في جميع أنحاء العالم تنفق مبالغ هائلة من المال لتطوير أجهزة محاكاة الطيران التي يمكن أن تحاكي الحياة الواقعية عل قدر الامكان ، على سبيل المثال في عام 2012 أنفقت (Microsoft) 4.2 مليونا لتطوير لعبة (flight x simulator) إلى جهاز محاكاة طيران حقيقي لتدريب الطيارين علي اساسيات الديناميكا الهوائية للطائرات. يحاول العديد من صانعي الطائرات العملاقة تطوير نظام طيران تلقائي (AFS) يمكنه حقًا أن يطير الطائرة من طرف إلى آخر ، لأن الطيار الآلي الحالي يمكنه فقط القيام بكمية محدودة من المهام ويمكنه فقط الطيران في ظروف محدودة بسبب الطاقة والتقنيات الحسابية المحدودة ، ولكن التعلم الآلي في الوقت الحالي يتمتع بشعبية وقوة كبيرة بحيث يمكننا أخيرًا تحقيق هذا الحلم من خلال تدريب نموذج على قيادة طائرة دون الحاجة إلى أي تفاعل بشري
Student Care CS406x
(MSA University, 2019) Adel, Monika
Education is the way of transferring knowledge from generation to another to keep life of human on earth, it is basic life need as food & shelter, now a day with the modern life need o understand all sciences and cope with it need more help and support beside the school learning which is one of the catastrophic problem of each home have children learning in schools is finding the right help & support. What about if there is transfer from current home residence to another in Egypt, how to manage finding new school teachers who can help for the Private Tutoring. This is the reason for reaching to an Application that has two interfaces the user who's is that the folks of the scholars and another one is that the teacher. This app plan is concerning the fogeys of scholars UN agency wants facilitate or additional care in their studies so that they join up during this app and search by place or space and therefore the subject they have additional care in it to search out the matching teacher. The user can notice an inventory of academics and teacher with their ratings and account that is his full name the themes he is specialized in it and wherever he provides the non-public categories The parent is ready to put in writing what percentage youngsters he has and therefore their age and what categories and subjects they take and he is the one UN agency writes the ratings concerning the doctors and academics the positives and the negatives. If you think it is just a “teaching Facebook”, you are wrong. This app allows parents and teachers to communicate and collaborate by sending messages, setting calendar reminders and much more. This application is one of many of the website that takes you on a journey across the whole platform so you can see how people use this social media so this application is going to take you in a trip with the teachers in every subject.
Detection of cervical cancer using computational genome assembly
(October University of Modern Sciences and Arts, 2019) Abo-Ouf Hamed, Yasmeen
Cervix cancer is one of the deadliest types of cancer in females because it can spread to other parts in the body damaging them as well. High‑risk Human PapillomaVirus (HPV) infection is a major cause of cervical cancer. Early detection is one of the most effective methods to combat cervical cancer. It increases the chances of survival drastically for the patient. High-throughput sequencing technology is the state of the art techniques for classification problems. With the development of Next Generation Sequencing (NGS) technologies, DNA sequencing (DNA‑Seq) has become a powerful tool for analyzing cancer transcriptomes, detecting such items as alternative splicing, isoform usage, gene fusions and novel transcripts with greater accuracy and higher efficiency. This thesis reported transcriptome sequencing analysis of two hundred and ten cervical cancer tissues and forty-three healthy cervical tissues. All the two hundred and fifty-three samples examined with before and after trim to produced quality reports, to ensure the quality of the data. Paired-end fastq files mapped with the reference genome, which is Human PapillomaVirus (HPV). Through applying backward search with Burrows-Wheeler Transform (BWT) algorithm, to efficiently align short sequencing reads against a long reference sequence such as the human genome, allowing mismatches and gaps. The thesis provides a transcriptome landscape of cervical cancer in Japanese, and Mexican patients and an improved understanding of the genetic regulatory network in HPV gene annotation. The accuracy is various as that the accuracy for Mexicans is from 10% to 50%, and Japanese is from 50% to 97%. This thesis tries to detect cancer before it turns into nodule as early as possible. : يعد سرطان عنق الرحم أحد أكثر أنواع السرطان فتكًا في الإناث لأنه يمكن أن ينتشر إلى أجزاء أخرى من الجسم يضر بها أيضًا. تعد عدوى فيروس الورم الحليمي البشري عالية المخاطر (HPV) أحد الأسباب الرئيسية لسرطان عنق الرحم. الاكتشاف المبكر هو أحد أكثر الوسائل فعالية لمكافحة سرطان عنق الرحم. لأنه يزيد من فرص البقاء بشكل كبير للمريض. تقنية التسلسل عالية الإنتاجية هي أحدث التقنيات لمشاكل التصنيف. مع تطور تقنيات تسلسل الجيل التالي (NGS) ، أصبح تسلسل الحمض النووي (DNA q Seq) أداة قوية لتحليل نسخ السرطان ، واكتشاف عناصر مثل الربط البديل ، واستخدام الإسوفورم ، والانصهار الجيني ، والنصوص الجديدة بدقة أكبر وكفاءة أعلى . ذكرت هذه الرسالة تحليل تسلسل النسخ من مائتين وعشرة أنسجة سرطان عنق الرحم وثلاثة وأربعين الأنسجة عنق الرحم صحية. جميع العينات مائتان وثلاثة وخمسين فحصها قبل وبعد القطع لتقارير الجودة المنتجة ، لضمان جودة البيانات. ملفات fastq المزدوجة نهاية المعينة مع الجينوم المرجعي ، وهو فيروس الورم الحليمي البشري (HPV). من خلال تطبيق البحث المتخلف باستخدام خوارزمية Burrows-Wheeler Transform (BWT) ، لمحاذاة التسلسل القصير يقرأ بفعالية مقابل تسلسل مرجعي طويل مثل الجينوم البشري ، مما يتيح عدم التطابق والثغرات. تقدم أطروحة المشهد نسخة من سرطان عنق الرحم في المرضى اليابانيين والمكسيكيين وفهم أفضل للشبكة التنظيمية الجينية في الشرح الجيني فيروس الورم الحليمي البشري. الدقة مختلفة حيث أن دقة المكسيكيين تتراوح من 10٪ إلى 50٪ ،
Self-driving car
(October University for Modern Sciences and Arts, 2019) Sayed Mahmoud, Muhammed
In this paper we will explain what an autonomous vehicle is and how the idea started and evolved through the 20th century to our modern times. Autonomous cars have the potential to save lives because most of the accidents are based on human error, not only that but also it has the potential to save fuel because they always drive the car in optimal condition. Moving on we explain what the different type of neural network architecture that are suitable for the self-driving car application. Then, we introduce the idea of using end-to-end deep learning in self-driving car, and how we used a 9-layer convolutional network model that takes augmented images and predicts a steering angle. Before the images are fed into the neural network it goes through augmentation process where we add artificial shifts and rotations. In testing we take two approaches the lidar based approach where we test the model, that we build which takes the lidar points as input, and the image-based approach where we use the convolutional network model mentioned above. في هذه الورقة سوف نوضح ماهية السيارة المستقلة وكيف بدأت الفكرة وتطورت خالل القرن العشرين حتى عصرنا الحديث. السيارات ذاتية الحكم لديها القدرة على إنقاذ األرواح ألن معظم الحوادث تعتمد على خطأ بشري، ليس هذا فحسب، بل أي ًضا لديها القدرة على توفير الوقود ألنها تقود السيارة دائ ًما في حالة ممتازة. مع التقدم، نوضح نوع مختلف من بنية الشبكة العصبية المناسبة لتطبيق السيارة ذاتية القيادة. بعد ذلك، نقدم فكرة استخدام التعلم العميق من البداية إلى النهاية في السيارة ذاتية القيادة، وكيف استخدمنا نموذ ًجا للشبكة التالفيفية المكونة من 9 طبقات والتي تلتقط صوراً معززة وتتوقع زاوية توجيه. قبل إدخال الصور في الشبكة العصبية، تمر عملية تكبير حيث نضيف تحوالت ودورات صناعية. في االختبار، نتخذ نهجين: النهج القائم على اللدائن حيث نقوم باختبار النموذج، والذي نقوم ببنائه والذي يأخذ نقاط اللدائن كمدخالت، والنهج القائم على الصور حيث نستخدم نموذج الشبكة التالفيفية المذكور أعاله
Smart Shopping Cart
(MSA University, 2019) Ibrahim Ahmed, Islam
Shopping at mall is a daily activity in our life. People buy different items at the malls and put them in the shopping carts. Here we can see the shopping cart is very helpful for costumers. Although the shopping cart is very helpful but it can make the customers very tired. And if they have a problem with their arm or leg or any kind of disability so this makes it difficult to use it. Here we come with an idea that makes a smart shopping cart follow the costumer by using a Bluetooth on android application to make it easy to use and to help people who got a problem or a disability. To solve this problem we can make the shopping cart work as tour guide to show us the place of the shopping items the customer needs. By using an application connected to GPS and Database system of the market. The shopping cart will also avoid obstacles automatically. This makes shopping much easier and fun.
Shopping and Delivery from obscure service providers
(MSA University, 2019) Gado, Mohannad
FindMe is a mobile application for finding services providers in certain locations based on the user’s location, which allows him to locate different services providers around him without having to cruise around to find them as their locations are not available on the google maps application. Those services may be patisserie shops, electricians, plumbers, café, vegetables and fruits markets and small supermarkets that are not pinned or couldn’t be found on google maps. FindMe will provide users with the locations of all the surrounding services providers so that it will be much easier to reach them in addition to giving the users the ability to evaluate those vendors and give them ratings based on their experience, so that the new users could make up their minds about which vendor should they deal with no need to try each one of them. Another benefit for this application is that it connects the users with riders in the area to let them fetch their orders from the vendors and deliver them to the users’ home in case that those shops don’t have the delivery service.
Detecting Depression from Text Using Convolutional Neural Networks and Word Embeddings
(October University of Modern Sciences and Arts, 2019) Amr Samir, Kareem
Depression has proved to be a major issue in our world for several decades. History teaches us that almost everyone is at peril of suffering from depression. Due to late diagnosis or none at all, millions of people unknowingly suffer from depression which then leads to catastrophic results. Previous studies have shown promise in detecting depression cues from text using Natural Language Processing techniques, using data compiled and manually annotated from Twitter. This project proposes to use a compiled list of manually annotated Twitter users as depressed or not, which was then used to retrieve said users’ tweets. Moreover, given that Convolutional Neural Networks have been shown to be quite versatile in the problem of text classification, especially when coupled word embeddings as features – this was used the classification model to solve our problem. : لقد ثبت أن الاكتئاب يمثل مشكلة رئيسية في عالمنا لعدة عقود. يعلمنا التاريخ أن الجميع تقريبا معرضون لخطر المعاناة من الاكتئاب. ملايين الأشخاص يعانون من اكتئاب بسبب التشخيص المتأخر أو لا شيء على الإطلاق، مما يؤدي إلى نتائج كارثية. لقد أظهرت الدراسات السابقة واعدة في اكتشاف إشارات الاكتئاب من النص باستخدام تقنيات معالجة اللغة الطبيعية ، وذلك باستخدام البيانات التي تم تجميعها والتعليق عليها يدويًا من Twitter. يقترح هذا المشروع استخدام قائمة مجمّعة من مستخدمي Twitter المشروحين يدويًا على أنهم مكتئبون أم لا ، والتي تم استخدامها بعد ذلك لاسترداد تغريدات المستخدمين المذكورة. علاوة على ذلك ، نظرًا إلى أن الشبكات العصبية التلافيفية قد أثبتت أنها متعددة الجوانب تمامًا في مشكلة تصنيف النص ، خاصةً عند استخدام حروف الكلمات المزدوجة كميزات - تم استخدام نموذج التصنيف هذا لحل مشكلتنا.
Colorization Using Deep Convolutional Neural Networks
(October University of Modern Sciences and Arts, 2019) Khaled Darwish, Sherine
يتم تقديم تقنية عامة لتحويل الصور ذات التدرج الرمادي إلى صورة ملونة ، ولتحقيق ذلك ، يلزم استخدام تقنية معينة للتنبؤ ببعض لون بكسل معين في الصورة بشكل تكيفي باستخدام أساليب التعلم العميق. على الرغم من أن إضافة اللون إلى الصور ذات اللون الرمادي يمكن أن يساعد في تحسين كل من المظهر البصري والتعبير. تم استكشاف بنى الشبكة المختلفة والأهداف ومساحات الألوان وتركيبات المشكلات. تسمى الطريقة المستخدمة في هذه الورقة الشبكة العصبية التلافيفية العميقة حيث لعبت دورًا بارزًا في معالجة الصور ورؤية الكمبيوتر. يمكننا تدريب صورنا للحصول على صور ملونة تتكون من 3 قنوات. يتدرب المشروع في الغالب على فراغ لوني مختلف عن كل الصور ، حيث يتم تحويل كل صورة إلى فراغ لوني L * a * b * ويمر حسب النموذج الذي تم إنشاؤه. ينصب التركيز الرئيسي في هذا المشروع على نموذجين مختلفين ، أحدهما لكل قناة متوقعة "a" و "b". يمكن أن تتغير خسارة القنوات المتوقعة من فئة إلى فئة أخرى. يمكن استخدام هذه الفكرة نفسها في تحويل مقاطع الفيديو ذات التدرج الرمادي أو تلك القديمة إلى مقاطع ملونة مثل الأفلام القديمة ، ويمكن أن يساعد تلوين الفيديو كثيرًا في الأفلام وتسجيلات CCTV القديمة التي تسجل بالأبيض والأسود. قم بتغذية النموذج بمجموعة من الصور التي يمكن تحويلها إلى مشهد ملون نهائي. يمكن أن يساعد تلوين الفيديو كثيرًا في الأفلام وتسجيلات CCTV القديمة التي تسجل بالأبيض والأسود. قد يكون تحسين النموذج في المستقبل من خلال تغذية النموذج بمجموعة من الصور التي يمكن تحويلها إلى مشهد ملون نهائي.
Gender Recognition Using Euclidean Distance Classification
(October university for modern sciences and arts, 2019) Osama Khairy El-Tamimi, Ibrahim
Gender recognition plays an important role in many fields, where many industries want to provide this service of gender automatic recognition. It also plays a big role in marketing strategies where this system can help tell which gender is interested in a specific product more. This project will be able to identify whether a person in a specific image is male or female using SIFT descriptor key points, the feature vector of the faces from images, classify the input using the Euclidean distance and the success percentage after using several testing methodology and the best percentage came from one to average method which is 70%. For future work of this project, Machine learning could be added and will change a lot in the final results as in using machine, the key points that being generated could be used in training several images and by testing it will automatically give me the result whether the gender is female or male without using computational method and a lot of mathematics as in Euclidean distance method. التعرف على نوع الجنس له دورًا مهمًا في العديد من المجالات ، حيث ترغب العديد من الصناعات في تقديم هذه الخدمة للتعرف التلقائي على النوع الاجتماعي. كما أنه يلعب دورًا كبيرًا في استراتيجيات التسويق حيث يمكن لهذا النظام المساعدة في تحديد النوع الاجتماعي المهتم بمنتج معين أكثر. سيكون هذا المشروع قادرًا على تحديد ما إذا كان شخص ما في صورة معينة ذكرًا أو أنثى باستخدام نقاط مفتاح واصف SIFT ، وموجه المعالم للوجوه من الصور ، وتصنيف المدخلات باستخدام المسافة الإقليدية ونسبة النجاح بعد استخدام منهجية الاختبار و جاءت أفضل نسبة من طريقة واحدة إلى متوسطة وهي 70٪. بالنسبة للعمل المستقبلي لهذا المشروع ، يمكن إضافة "التعلم الآلي" وسوف يتغير كثيرًا في النتائج النهائية كما هو الحال في استخدام "الجهاز" ، ويمكن استخدام النقاط الرئيسية التي يتم إنشاؤها في تدريب عدة صور ومن خلال اختباره ، سيعطيني تلقائيًا النتيجة سواء الجنس أنثى أو ذكر باستخدام الطريقة الحسابية والكثير من الرياضيات كما هو الحال في طريقة المسافة الإقليدية.

Browse

Browsing Faculty Of Computer Science Graduation Project by Issue Date