VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT PROJECT NAME: Research on CNN applied in face recognition of International School students Student’s name: Le Thuy Hang Hanoi, 2023 VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT PROJECT NAME: Research on CNN applied in face recognition of International School students SUPERVISOR: Ph.D Le Trung Thanh STUDENT: Le Thuy Hang STUDENT ID: 18071520 MAJOR: Informatics and Computer Engineering CLASS: ICE2018A Hanoi, 2023 2 LETTER OF DECLARATION I hereby confirm that my research led to the creation of the graduate project, "Research on CNN applied in face recognition of International School students" and that this work hasn't been published before. The topic of the CNN applied in face recognition was developed based on the topic of Ph.D Le Trung Thanh. I take the principles of research ethics very seriously while I carry out my endeavor. My own research and surveys served as the basis for all project findings, and all sources were correctly referenced.
I accept complete responsibility for the correctness of the data, statistics, and other papers in my graduation project. Hanoi, 30 May 2023 Student Le Thuy Hang 3 ACKNOWLEDGEMENT To be able to conplete this research, I would like to take the opportunity to thank Ph. Le Trung Thanh for mentoring me as I developed my final project. He provided me with all the advice and support I needed, as well as all the resources I needed to do my work more quickly.
He also shared me some advice that I might use. It is really difficult for me to complete my final assignment without the teacher's help. However, given my lack of experience, I am sure that I will make blunders. As a result, I am looking forward to the professors' input to help us finish the project and create a firm foundation of knowledge that will help us reshape ourselves in the future.
Besides, i would like to sincerely thank my family, brothers for their enthusiatic contribution and sharing experiences to help me finish this research. Again, I want to take this opportunity to once more thank all of my professors, friends, and family for their unwavering support and dedication to helping me become a better person. 4 Table of Contents LETTER OF DECLARATION. 9 Chapter 1: Introduction to the topic.
The topic’s scientific and practical importance. What is the goal of the topic. Why was this topic chosen?. 12 Chapter II: Theoretical basic .1 Biological neural network.2 Artificial neural network [4] .3 Deep learning network model .1 Some Deep Learning Network Components.
15 Chapter III: Design and build models .1 Introduction to Matlab .2 Create train and test datasets. 44 Chapter 4: Results and Discussion .1 Model evaluation results.2 Discuss model accuracy and performance. 49 Chapter V: Conclusion and development direction. 56 5 Table of Figure Figure 1 : Biological neural network.
13 Figure 2: Humans describe the processing architecture at a neuron. 13 Figure 3: CNN Model Network. 14 Figure 4: Convolutional neural network operational structure. 15 Figure 5: Describe the convolution layer.
17 Figure 6: Convolution X filter = Feature Map. 18 Figure 7: Polling layer. 18 Figure 8: Full connected layer. 19 Figure 9: Sigmoid function.
20 Figure 10: Tanh Function. 21 Figure 11: ReLU function. 22 Figure 12: Activation softmax. 23 Figure 13: CNN feature layers.
24 Figure 14: Convnet Configuration. 25 Figure 15: : The value of the log loss when the accuracy rate reaches 1. 27 Figure 16: Network structure Alexnet. 28 Figure 17: Dropout technique.
36 Figure 19: The Hangle dataset after being collected. 37 Figure 20: Original size photo from collection. 41 Figure 21: Original image after the surrounding parts have been removed. 42 Figure 22: The face in the picture is different when the surrounding is cut off.
42 Figure 23: The image after it has been resized to the size 224x224. 43 Figure 24: The image of the dataset after it has been split to the training and validation part. 44 Figure 25: Using Data Augmentation technique. 45 Figure 26: Alexnet model parameters.
46 Figure 27: Training results when setting inititalLearnRate 0. 47 Figure 28: Training results when setting inititalLearnRate 0. 47 Figure 29: Results after training. 48 Figure 30: The results of the predictive model of the True Position (True Class).
49 Figure 31: Other image prediction results in the dataset. 49 Figure 32: Face prediction results through camera. 50 Figure 33: Photo prediction results from another device. 50 6 Abbreviated keyword table ID Acronyms Full word 1 ANN Artificial neural network 2 CNN Convolutional Neural Network 3 ReLU Rectified linear unit 4 RNN Recurrent Neural Network 5 GAN Generative Adversarial Network 6 PPV Positive Predictive Value 7 List of mathematical formulas Formula 1: Neural network.
14 Formula 2: Calculator tanh function. 21 Formula 3: Calculator Relu function. 22 Formula 4: Calculator softmax function. 23 Formula 5: Calculator Cross Entropy.
27 Formula 6: calculator regularization loss. 31 8 Abstract Nowaday, Student management is one of the basic and important issues of universities. However, in large schools, student management can be challenging, especially in controlling student identities. Artificial intelligence is one of the strongly developed technologies, which can be applied in solving problems in daily life.
In the field of education, applying artificial intelligence to solve problems in student management is becoming a new and advanced trend. The project "Research on convolutional neural networks ” when identifying international school students at Vietnam National University, Hanoi seeks to develop an applied convolutional neural network (CNN) model for recognizing international students at international schools, Vietnam National University, Hanoi. This research has major implications for managing students at the international school in the actual world as well as for expanding our understanding of model training and image processing approaches. Student administration will become more simple and straightforward, while also taking less time and money thanks to the use of a convolutional neural network to identify students in images, while improving the quality of education.
From investigating the convolutional neural network in student identification, we may broaden the topic to examine the recognition of other things such as faces, license plates, items, and so on. This will aid in increasing the application of convolutional neural networks in a variety of industries while also improving the usefulness and growth of artificial intelligence technologies. In summary, the research content will revolve around the application of convolutional neural networks to identify students on photographs, including steps to build datasets, design and train models, evaluate and compare results. results with existing traditional methods.
9 Chapter 1: Introduction to the topic What is the scientific and practical significance of the topic, what is the goal of the topic? the inspiration for choosing the topic and research content will be mentioned in this chapter. The topic’s scientific and practical importance The topic "Research on convolutional neural networks applied in face recognition of international school students at Vietnam National University, Hanoi " is important both scientifically and practically. In terms of science, the topic gives a detailed research on the application of convolutional neural networks to the problem of image identification of students. This contributes to the study and development of techniques and technologies connected [1], There are many applications for artificial intelligence technology, and it is growing quickly.
The development of methods to tackle issues in this area is aided by this study, which also advances knowledge of model training and image processing techniques. In terms of practice, the topic has great significance in solving practical problems related to student management at the international school of Vietnam National University, Hanoi. The application of a convolutional neural network to identify students on photos will make student management easier and more convenient, and at the same time reduce the time and cost of student management procedures. This research can also be applied to other schools and organizations, helping to reduce the time and cost of the student management process, while improving the quality of education.
What is the goal of the topic The project "Research on CNN applied in international school student identification at Vietnam National University” intends to create a convolutional neural 10 network (CNN) model to identify international school students at Vietnam National University in Hanoi. This study will specifically determine the following goals: 1. Collect student image data from different sources and preprocess the data to prepare for training the CNN model. Train the CNN model on student image data and evaluate the model's effectiveness.
Apply the trained CNN model to identify actual students and evaluate the accuracy of the model in recognition. Compare the effectiveness of the CNN model with traditional identification methods to evaluate the contribution of this research. Why was this topic chosen? The project " Research on CNN applied in face recognition of International school students " was selected as my graduation project based on the following highlights: [3]. Applying artificial intelligence to challenges in student management is becoming into an exciting and modern trend in education.
One of the applications of artificial intelligence in student management is using a convolutional neural network (CNN) to identify students through photos. CNN is one of the types of neural networks widely used in image and audio processing. It is capable of learning image features and using them to classify different objects in the image. Therefore, the topic "Research on convolutional neural networks applied in international school student, Vietnam National University, Hanoi" is a very relevant and highly applicable topic.
This research can make it easier to manage students at the school and save a lot of time and money. The implementation of this thesis will help me learn and develop skills in the field of artificial intelligence and image processing, two fields that are very developing and promising in the future. In addition, this topic is highly applicable and makes a meaningful contribution to the Vietnam National University, Hanoi. Research methods The content of the research "Research on convolutional neural networks applied in international school student of Vietnam National University" will be divided into the following main parts: 1.
Overview of Convolutional Neural Networks 2. Methods of identifying students on photos 3. Building a student identity dataset 4. Design and train convolutional neural models 5.
Evaluate and compare results 6. Discussion and conclusion 1.5 Project layout The layout of the report is divided into five main chapters as follows: Chapter 1: An overview about topic Chapter 2: Theoretical basic Chapter 3: System design and model building Chapter 4: Implementation results and discussion Chapter 5: Conclusion and development direction In conclusion, this chapter provided an overview of the topic, It can be said that the very fast development speed has shown the superiority of the application of information technology in educational management. This is the reason why the topic for the graduation project was picked and then learn about the best way to complete this topic. 12 Chapter II: Theoretical basic In this chapter, we offer methods to solve problems in the problem of recognizing violence.
Definitions, theoretical basis as well as advantages and disadvantages of the method will be presented in detail with some examples and descriptive images.1 Biological neural network Figure 1 : Biological neural network Biological neural networks are made up of a collection of chemically or functionally related neurons. The overall number of neurons and connections in a network may be large, and a single neuron may be linked to many different neurons. Synapses, which can also be dendrodendritic synapses and other connections, are often created when axons and dendrites come together. Apart from electrical signalling, there are other forms of signalling that arise from neurotransmitter diffusion [1] [2].