VIETNAM NATIONAL UNIVERSITY THE UNIVERSITY OF INFORMATION TECHNOLOGY THESIS REPORT AN INFORMATION SYSTEM FOR THE CONSERVATION OF VIETNAM’S BIODIVERSITY. Nguyen Thanh Binh Mai Van Binh - 17520280 Nguyen Thanh Duc - 17521296 Ho Chi Minh City, 2021 VIETNAM NATIONAL UNIVERSITY THE UNIVERSITY OF INFORMATION TECHNOLOGY FACULTY OF INFORMATION SYSTEMS Vy“ THESIS REPORT AN INFORMATION SYSTEM FOR THE CONSERVATION OF VIETNAM’S BIODIVERSITY. Nguyen Thanh Binh Mai Van Binh - 17520280 Nguyen Thanh Due - 17521296 Ho Chi Minh City, 2021 ASSESSMENT COMMITTEE The Assessment Committee is established under the Decision. , date -by Rector of the University of Information Technology.
— ieee cee eee eee eeeee een eeeeaeeeeneenes — Chairman.=— e eee e eee e tesa eee eeaeeee esa eea esa eeae ee — Secretary. Boece eee — Member. ii iii ACKNOWLEDGEMENTS We would like to thank Dr. Nguyen Thanh Binh, our thesis director, for his time, excellent assistance, insightful reviews, and invaluable advice.
This job would not have been possible without his dedication to our success. He is a continual source of motivation for us and assists us in honing our talents, which came in handy when we were working on our thesis. We have a great deal of regard for him in my heart. Weare grateful to both our supervisor and my university, particularly the Faculty of Information Systems, in addition to our supervisor.
We would also like to express our gratitude to all the teachers who have taught us at The University of Information Technology for the past four years. They have made it their mission to teach their students, including us. As a result, we hold them in high regard. Mai Van Binh & Nguyen Thanh Duc — students of AEP 2017.
iv TABLE OF CONTENTS ASSESSMENT COMMITTEE .ccsccscssssssssssssssssssesssssscnsssensssenensssensssesenseeenees I COMMENTS ON INSTRUCTORG.ccsssssssssssssessssesessessssssesessessseeseeesseseeeees II COMMENTS ON REVIEWERG.ccccsssssssssessssesessssessssssesssesnscssessseesnsseees II ACKNOWLEDGEMENTS TABLE OF CONTENTS. LIST OF FIGURES .ccccsssssssssssssssssssssscsssssssssssssesssessssscssscessesssseessenseeeses VINH LIST OF TABLES .--- 5555555 2sSESsSESESESErkerekerrerrasnsrsnnsrae X LIST OF SYMBOLS AND ABBREVIATIONS. co OT cỈnHHneiiene XII CHAPTER 1. Status of research.
Scope of project 1. 6-11 về ngư 2 CHAPTER 2. Creatures’ classification SySteM. Redbook classification syStem.
Standards to assess the states of species in Viet Nam red data book. Standards to assess the states of species in IUCN Red List (The IUCN Red List of Threatened Animals). Endemic species classification system. Framework, platform, and technology.
Framework and platform. Creatures and plants image classification 2. Image classification overview 2. Creatures’ classification to flower classification problem.
Traditional machine learning methods. Deep learning methods. Convolutional Neural Network — CNN. Sàn HT HH iưy 20 3.
SQL data types preprocessing. HTML file data 3. Result of data collection Process. SYSTEM ANALYSIS AND DESIGN.
Modules and feature design. RESTful API architecture 4. Some functions in public website 4. Function to help beginners 4.
Questions and Answers. Identify and Contributions. Some functions in admin website 4. Questions and answers manageMent .--- - «+ ssx+e+exeeeeeerer 57 CHAPTER 5.
AUTOMATED FLOWER CLASSIFICATIO LARGE NUMBER OF CLASSES. Transfer learning approach. Image pre-processing and data augmentation. Flower classification base on transfer learning.
Flower classification base on Vision Transformer (ViT). Overall performance evaluation. Integrate into the system. DEPLOY AND TEST SYSTEM.
CONCLUSION AND FUTURE DEVELOPMENT DIRECTION T7 TA. Novelty and innOVafIV€. Summary of research results. Vision and ambition.ccccssscsssssssssssssssesssesssssscscsssescsssesesssesssesssnsesasscassessscssscensseeseseees 79 Vii Figure 2-1.
Genus of orangutan. classification system used to identify animals. Rules on states of threatened species and the categorization of them Figure 2-4. Filter of Viet Nam endemic species.
Framework, platform, and technology. Code HTML Figure 2-7. Add CSS into HTML code. The challenges involved from the perspective of a Computer Vision algorithm Figure 2-10.
An example training set for four visual categories. Inter-class similarity (Osteospermum, Pericallis, and Gazaniarigens) Figure 2-12. Intra-class variability (view difference). CNN image classification pipeline.
Architecture of a CNN. Original database Figure 3-2. SQL data types processing steps. The new ERD schema.
Example of html file data. Extract information about organism’s html page. Extract information about Redbook creature’s html pag Figure 3-7. Post html page example.
Interface of footprint has Viet Nam name, Latin name, image, and id of creatures Figure 3-9. Interface of timber samples has Viet Nam name, Latin name, image, and id of creatures Figure 3-10. Corrected data entity model. The function modules of the system Figure 4-4.
The entity-relationship model. RESTful API architecture. Contributor for my system. Characteristics of species.
Specialized knowledge for users. Timeber samples screen Figure 4-15. Latin-Viet dictionary. Post's detail screen.
Specific creature interface. Level ofred list rating Figure 4-21. National park: Figure 4-23. National Park detail.
Questions and answers. Animal protection campaign Figure 4-28. Event screen for user Figure 4-29. Register form Figure 4-31.
Register event diagram. Identify plants UI. Response for use’s contribution. user can validate propose name and decide to contribute or not 51 Figure 4-35.
List of user’s contributions. Post management Figure 4-37. Screen manager events. Event creation screen Figure 4-40.
Event participants information. Question management screen. Question preview and management answers Figure 4-43. Create question diagram Figure 4-44.
Create answer diagram. Contributions management screen Figure 4-46. Contribution activity diagram Figure 5-1. A sample of the flower images Figure 5-2.
Image pre-processing and data augmentation Figure 5-4. Standard ConvNet Concept. A 5-layer dense block. A deep DenseNet with three dense blocks.
The layers between two adjacent blocks are referred to as transition layers and change Figure 5-7. DenseNet architectures for ImageNet. ViT Model Overview. The intuition of Linear Projection Block before feeding in Encoder.
N is the number of patches form the image Figure 5-11. Positional Embeddings process. Transformer Encoder Architecture Figure 5-13. Self-Attention layer.
Self-attention layer works. Multi-head Attention. MLP Head results Figure 5-17. Plots loss on training and validation sets over training epoch: Figure 5-18.
Flower classification Server-Client architecture. Add a image for classification task. Results for flower classification task. ix LIST OF TABLES Table 4-1.
Tables list information. Performance of proposed methods LIST OF SYMBOLS AND ABBREVIATIONS Abbreviations Explanation HTML Hypertext Markup Language CSS Cascading Style Sheets JSX JavaScript Syntax Extension CNNs Convolutional neural networks DCNN Deep convolutional neural network ReLU Rectified Linear Unit API Application Programming Interface ViT Vision Transformer XI ABSTRACT Currently, in Vietnam, there are many diverse species of organisms, but very few people can clearly understand the species that are distributed and developed in Vietnam. Therefore, in order to help Vietnamese people better understand biodiversity, we decided to collect data and build a website to serve the organism search community so that people can easily find information about flora and fauna and can exchange information with each other through the forum page. In addition, there will be an admin page to support management, update information about newly found creatures or edit information.
The project will use the power of deep learning to build the power search engine. Introduction Have we ever wondered what the natural flora and fauna of Vietnam is like, rich or poor, current status and distribution? Our research topic was inspired by this question. We want to help researchers have a standard reference library on Vietnam's biodiversity, want to give the community answers about the current state of nature in Vietnam, and want to bring love for nature and the spirit of mother protection, as natural as human survival goals for children. Although Vietnam has only a modest area, but due to the great difference in climate from the region near the Equator to the subtropical region, along with the diversity of topography has created a diversity of nature.
And because of that, Vietnam has a very high biodiversity. Vietnam's wildlife includes 11,400 species of vascular plants, 1030 species of mosses, 310 species of mammals, 390 species of reptiles, 262 species of amphibians, 700 species of freshwater fish and 2,000 species of marine fish. Vietnam is also home to thousands of species of plants, mosses, and fungi. The team wants to create a plant and animal information system where you can find all the information you need about any animal that occurs in the territory of Vietnam.
The system will be a reliable source of information that can be used in other scientific research topics and projects. Have you ever been confused when your child or grandchild asks about the name of an animal in a zoo or in the forests of Vietnam when you have a chance to take your child or grandchild there, but you cannot answer or answer the story completely? If you are a biology teacher, are you confident enough to properly introduce your students to Vietnam's Forest Creatures? If you answer incorrectly and answer is not confident enough, either intentionally or unintentionally, it means that you have inadvertently taught your children knowledge that you have not understood correctly. Then they will talk to their classmates again about the unwitting mistake you did for them, which is dangerous to the young minds of young children as they begin to explore the wild. With this scientific research topic, we will also provide people with a system so that people of all ages can have more accurate understandings of Vietnam's forest biome in an approach that brings entertainment.
You can ask questions with your children and grandchildren so that our system, where there are documents on biodiversity, helps you understand better and better about the things you do not know or do not have time to learn. In addition, to serve the needs of more understanding of all people, we want to create a place where people can easily register with protected areas, national parks, organize picnics, study extracurricular activities. Motivation The team consists of people who have a love for the country's nature. But the more we care about nature, the more we worry about how people are treating it.
That's why we decided to do something to help those who want to contribute to protecting nature will know about the places that are organizing volunteer activities to protect nature and rescue Vietnamese creatures to participate family. From there, the system can create a large community where everyone can participate and discuss volunteering campaigns taking place across the country. Status of research In the world, there are many websites about creatures, but only specialize in one species such as fish or butterfly. Example, Pl@ntNet website is only about plant or www.com, which is website in Viet Nam is the same.
The WCS VIETNAM only contains information about wild animals. Until now, there is no website that provides information about creatures in Vietnam as much as we do. It can be said that we are currently the leading and almost the only one engaged in this field. Until now, there is no website that provides information about creatures in Vietnam as much as we do.
It can be said that we are currently the leading and almost the only one engaged in this field. Scope of project - Analyze, design, and build a practical information system for the purpose of conservation, research, and community education on biodiversity in Vietnam. - The system provides a reliable source of data for users to be able to look up information on species in Vietnam, protected areas, national parks, and nature conservation organizations under the main auspices. - The source of data, besides the accurate information of high academic science, also has common knowledge about nature, vividly illustrated, highly entertaining animals aimed at children.
Primary school, junior high school students and ordinary people love and want to learn about the vivid nature of Vietnam. - The data is designed with high association between species, families, orders, protected areas, national forests, conservation organizations, ecological environment, geographical features, environment, and social and spiritual characteristics. Contributions We receive a commitment to support official content from freelance biodiversity researcher Phung My Trung. We also have been provided with datasets from national parks such as Vietnamese national parks such as Nui Chua National Park, Nam Cat Tien National Park.