NATIONAL UNIVERSITY OF HO CHI MINH CITY UNIVERSITY OF INFORMATION TECHNOLOGY FACULTY OF COMPUTER ENGINEERING BANH NGOC NINH DAO DUY KHANG GRADUATION THESIS SMART-TABLES ON IOT PLATFORM, APPLIED IN SMART CLASS BAN HỌC THONG MINH TREN NEN TANG IOT, UNG DUNG TRONG LOP HOC THONG MINH ENGINEER IN COMPUTER ENGINEERING HO CHI MINH CITY, 2021 NATIONAL UNIVERSITY OF HO CHI MINH CITY UNIVERSITY OF INFORMATION TECHNOLOGY FACULTY OF COMPUTER ENGINEERING BANH NGOC NINH- 17520869 DAO DUY KHANG- 16521697 GRADUATION THESIS SMART-TABLES ON IOT PLATFORM, APPLIED IN SMART CLASS BAN HOC THONG MINH TREN NEN TANG IOT, UNG DUNG TRONG LOP HOC THONG MINH ENGINEER IN COMPUTER ENGINEERING INSTRUCTOR PhD PHAM QUOC HUNG HO CHI MINH CITY, 2021 INFORMATION OF THE GRADUATION COURSE GRADING BOARD The board marks the graduation course, established under Decision No.462/QD- DHCNTT, 23 July 2021 of the President of the University of Information Technology. THANKS First of all, we would like to sincerely thank the Board of Directors, teachers of the University of Information Technology, especially teachers in the Faculty of Computer Engineering who have dedicatedly taught and communicated valuable knowledge and experiences to us during the past time. With the knowledge taught by teachers in the Department of Computer Engineering for the past 4 years, the group has been able to implement this topic with a variety of knowledge along with new knowledge. Besides, to be able to perform well the topic is the guidance of Mr.
Pham Quoc Hung. We would like to express our deep gratitude to Dr. Pham Quoc Hung — Faculty of Computer Engineering who has dedicated and helped us in the process of implementing the topic. Wishing teachers of computer engineering faculty — University of Information Technology always have a lot of health to accompany us on the academic path.
Thank you to the teachers who have been by our side and led our students from the first days of entering the university lecture hall to date. At the end of the word, once again, our sincere gratitude to our teachers! Wishing Mr. Pham Quoc Hung more success in teaching and healthy family! Ho Chi Minh City, June 5, 2021 Students perform Banh Ngoc Ninh Dao Duy Khang Faculty of Computer Engineering MTCL2017.1 TABLE OF CONTENTS Chapter 1. Facial recognition technology.
Research on smart table. In the WOrÏd. System evaluation method. Objectives ofthe topic.
Practical meaning of the fOpIC. Structure of the essay T€DOFẨ. What is smart CaSSTOOI. Benefits of smart classroom in an educational environmenI.
What is smart (able.-cc St nh tình re9 2. Technical model for management and interaction between teachers and students in smart ClaSSTOOMS. LH ngu HH Hư 1 2. ch tà HH 200 11g.
Fully connected laYT. How to choose parameters for CNN.- St TH HH TH HH it 4 2.- - - ¿+ + + SE+k2k£E£ESEEkEEEEEEEkrkrkekererkrkree 17 PnnH:. -c5 tt TH TH gưưn 17 2. Database and database sySf€Ims.
cà HH HH re 21 2. Client-server model. Structured Query Language. How does MySQL Work?.
API - Application Programming Interface 2. ——¬—QẦ——s BE th erreie27 2.à cà tt nhe 28 Phố. Frameworks for JaVaSCTIDI. Frontend fram€WOTKS.-- - ¿+ SE HH1 00101 1 tà30 2.
Edge Computing vs Cloud Computing. Advantages of Using Edge Computing. Cloud Computing occ es - 6 - Sàn gà 32 2. Benefits of Using Cloud Computing.
SYSTEM DESIGN ANALYSIS. Hardware Achitecture of Smart TiabÌe. KIT Development Environment. Technical model for management and interaction between teachers and students in smart CÏ4SSTOOITS.
Teacher side wc. Student side oo. Facial Recognition and login method. Main functions and flow of smart table websites „4l 3.
Main functions and flow of the website take the online. Automatic student at(endane. Identify the right students throughout their studies and take tests on the SYSUCM.EXPERIMENTS AND EVALUATIONG. Face recognition experiment 0.
Train model process. -- - St HT H1 H1HHu HH. Interactive experiments with databaSe€s. Student admissions procesS.
Get information from the database. The process of exams with the method of identity. CONCLUSION AND ORIENTED OF DEVELOPMENT. What we've dON€.
Things that haven't been done. Oriented developIm€ni(.58 PICTURE CATEGORY Figure 1.1: The CROSS2DESIGN Smart Table.- ¿+ 5£ S£9E2EE#E#EEE#EE SE 91 1111113 1xx 8 Figure 2.2: Infrastructure model of digital education.3: Kernel in CNN.4: Describe how each kernel works with input Image.5: The's concerned about the CNN .6: Overview of MTCNN's 3 p-net, r-net, o-net processing classes.7: CNN diagram of P-net.8: CNN diagram of R-Net .9: CNN diagram of O-Net.10: Anchor, Positive, and Negative examples .11: Example of a link between a Client and a SeTVeT.12: The contact diagram between backend and frontend .13: The web API system .14: Edge Computing vs Cloud Computing.1: Hardware achiteC(UT€. - - 5+ St ky 34 Figure 3. St nt ng HH H0.4: Technical model for management and interaction.5: A facial recognition overview diagram for logging into the system .6: A diagram of login system 0.7: A diagram of the main functions in the smart table system.8: A diagram of the main functions in the exam system.9: The system's automatic attendance pDFOCeSS.10: Student identity process diagrams throughout the learning process .1: Conducting train mOeÌS.- esses 5+ +*+k‡‡k£EvEkEkekEkrkrkrkekerrree 47 Figure 4.2: Face recognition of stUd€TI{S,.-- - ¿5-5-6 St St+t+ktekerrkrkereree 48 Figure 4.3: The point-of-list đatabase.4: Notice of successful attendance (manually).5: Student and teacher management database diagram .6: Student's personal information stored as JSON .7: Each student's DO ÍiÍ€.8: The process of identifying students during the test.9: Images of students when study or make test captured from the system 55 Figure 4.10: The main interaction interface with students.---- 5< +ccs«++ 56 TABLE CATEGORIES Table 4.1: Set of photos representing training data for facial recognition.2: Test results in bright and near conditions (400 lux and below 30 cm).3: Test results in sufficiently bright (400 lux) conditions and over Im .4: Test results in low-light & near conditions (200 lux & below 30 cm).5: Test results in low-light (200 lux) conditions and over 1m .6: Comparison between smart team table and smart table in the world.
56 CATEGORIES ACRONYMS FPS - Frame Per Second loT - Internet of Things API - Application Programming Interface CNN - Convolutional Neural Network MTCNN - Multi-task Cascaded Convolutional Networks Al - Artificial Intelligence SVM - Support Vector Machine DCNN - Deep convolutional neural network CPU - Central Processing Unit PCA - Principal Component Analysis PHP - Hypertext Preprocessor ASP - Active Server Pages HTML - Hypertext Markup Language SQL - Structured Query Language URL - Uniform Resource Locator RDBMS - Relational Database Management System ABSTRACT Through the learning process, currently in the world and in the country there are smart tables designed with touch surfaces, students interact directly with the surface, bringing entertainment advantages but no facial recognition as well as functions for learning here. Develop facial recognition systems and build models for management and interaction to support the learning and teaching process of teachers and students. While storing and managing the information ofteachers and students, identities will be saved on the cloud service provided by the database management system platform. The system will have functions such as (1 - Facial recognition; 2 - Student presence management by automatic check-in; 3 - Allowing students to download documents; 4 - Identify students throughout the learning process and during the online test).
To serve the learning process (mainly at the high school level), my team decided to create a device called a smart desk. The device makes it possible for teachers to (1 - Manage student presence in lessons through devices such as cameras and touchscreens; 2 - Check-in automatically students when logging into the system; 3 - Share class learning resources; 4 - Identify students throughout the learning process and during the online test) | Chapter 1: TOPIC OPENING 1. Facial recognition technology Today facial recognition technology [1] is very popular in life and is known as an important area in identity science due to the need for facial recognition applications in many areas such as economics, crime investigation, security,. Facial recognition technology is a type of biometric software that mapped an individual's facial features mathematically and stored the data as a faceprint.
Facial recognition algorithms have been developed quite widely with a variety of solutions and algorithms. However, it is possible to generalize this processing in 4 steps as follows: Locate the face in the photo. Search for facial features. Compare with photos in the data set.
Export to face recognition. For humans, facial recognition takes place regularly in life, people easily recognize faces as well as identify objects whenever encountering other objects. But in order for computers to be able to do the same thing as humans, it is necessary to solve a lot of high-level problems and appropriate processing algorithms to carry out each step of processing. Therefore, building a facial recognition system to support IoT systems will be difficult and complex.
Using facial recognition technology can bring a range of potential benefits, including: - There is no need to directly come into contact with the device for authentication (other contact-based biometric authentication techniques such as fingerprint scanners, may not work correctly if there is a stain on a person's hand). - Less processing is required than other biometric authentication techniques. - Easily integrate with existing security features. - Accuracy improves over time.
- Can be used to help automate authentication. In general, facial recognition technology has been widely applied today and maybe in the future, this technology will develop even more, usefully applied in modern technology life. Research on smart table 1. In the world With technology such as smartphones and tablets being rather ubiquitous, the CROSS2DESIGN Smart Table comes as a logical next step when it comes to technology integration.
Blending the aesthetics ofa functional table with the immersive experience of a tablet, the table can be utilized for work, school and entertainment purposes, and everywhere in between. The touch-sensitive top makes it easy to interact with the on-screen content, while the robotic leg system allows it to be angled into different orientations. This allows it to be used as a traditional table, a screen for presenting and even an easel for digital designing.1: The CROSS2DESIGN Smart Table The disadvantage of those smart tables is that it lacks of interaction between teachers and students and that teachers cannot manage whether their students are actually in the classroom or not. Therefore, my team wants to create a smart table product that can help identify faces as well as interacting such as: sharing classroom resources, taking screenshots of work with students' faces when time runs out.
Project's goal To serve the learning process (mainly at the high school level), my team decided to create a device called a smart table. The device makes it possible for teachers to manage student presence in lessons through devices that support the accompanying camera and touchscreen, share class learning resources if they are students of the class and enable smart submissions by taking screenshots of work with the student's face when time runs out. Along with that, my team wishes to build a technical model for management and interaction between teachers and students in smart classrooms 1. System evaluation method Based on the accuracy and reliability of the system, the identification time 4 reaches the required speed or not, the reliability is high or not.
The evaluation method through the demo will use the mold of my team member to identify the match and use the teacher's face in case of non-matching, then will send the document if it coincides and give a notification if it does not coincide. In the process of learning, the system will self-identify the object. If there is a test, the system will capture the student's work and face when time runs out. Recommended solutions Learn documents and knowledge related to Multi-task Cascaded Convolutional Networks (MTCNN) [9] for face detection, FaceNet [8] for face recognition and Jetson Nano devices.