1 of 91

Advanced Student Monitoring System for physical classes using pattern recognition and behavioral analysis

2023-227

2 of 91

Meet Our Team

IT20189594

Rathnayka R. K. A. R.

IT20200206

Mallawarachchi S.M.A.

IT20191788

Wijesiriwardana H.G.N.D

IT20122850

Perera S.S.A.

3 of 91

Mr. Jeewaka Perera

Supervisor

Our Supervisor

4 of 91

Introduction

The system will incorporate facial recognition technology and computer vision algorithms to identify students, detect emotions and behaviors, and analyze the data to provide feedback to lecturers on how to improve their teaching strategies.

5 of 91

Research Question

Taking attendance in a large classroom can be a tedious and time-consuming task for teachers.

Traditional methods of taking attendance, can be inaccurate and easily manipulated.

Difficult to understand student behavior in large classrooms.

Teachers can't evaluate their performance due to the lack of student feedback.

6 of 91

Objectives

Develop a reliable face recognition algorithm that accurately identifies students in a classroom setting.

Efficiently store and manage and prostudent attendance data.

Develop a user-friendly interface for teachers to access the attendance data and analyze student behavior.

Optimize system to use in a classroom setting, where multiple students may be present in the camera's field of view.

Security measures to ensure the privacy and security of the stored student data.

7 of 91

Research Components

Emotion and behavior capture

Monitoring Student's Concentration levels

Student Attendance Marking with Face Recognition

Enhance low resolution footage

8 of 91

Overall System Diagram

9 of 91

Work Breakdown Chart

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

10 of 91

Camera position

option A

option B

11 of 91

Completion of the project

12 of 91

Commercialization

Improve and optimize the system larger class rooms and other potential markets such as educational institutions, corporate offices, and government agencies

Desktop app integration

Cloud based service, with a subscription based or one time payment options

13 of 91

Poster

14 of 91

Research Paper Acceptance

15 of 91

IT20191788 Wijesiriwardana H.G.N.D

Specialization : Information Technology

16 of 91

Enhance Low Resolution Footage

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

17 of 91

Background

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Student monitoring systems, the quality of video footage plays a major role in ensuring accurate analysis and decision-making

However, the implementation of high-quality cameras in real-world classroom settings often faces significant challenges due to budget constraint

Therefore we need a cost-effective solution that can enhance the visual quality of low-resolution footage

18 of 91

Research Gap

Research Paper [1]

Research Paper [2]

Research Paper [3]

Research Paper [4]

Research Paper [5]

Proposed System

SRGAN Method

Video Upscaling Enhancement

Camera Usage

Implement In Physical Classroom Environment

Integrate with Student Monitoring System

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

19 of 91

Research Problem

How to enhance low resolution footage with image upscaling

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

With the normal footage student faces may be not clear. If we enhance the video it helps to accurate analysis of our system.

20 of 91

Novelty

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Cost-Effective Enhancement: In educational settings, budget constraints often limit the high-quality cameras. Our component offers a cost-effective solution by improving the visual quality of existing low-resolution footage.

Contribution to Educational Technology: By enhancing low-resolution video, our research contributes to the field of educational technology, demonstrating the feasibility and effectiveness of improving student monitoring systems with advanced computer vision techniques.

21 of 91

Specific and sub objectives

Develop an innovative machine learning-based solution that can enhance low-resolution footage, significantly improving the scaling

Implement and integrate the SRGAN model into the student monitoring system to enhance the visual quality of low-resolution footage

Due to budget constraint of using high resolution cameras, we need a cost-effective solution that enhances the visual quality of low-resolution footage

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Enhance the video for better analysis.

22 of 91

System Diagram

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

23 of 91

Model Architecture

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

24 of 91

Integrated development environment (IDE)

    • PyCharm
    • Python
    • PyTorch
    • cv2 library by OpenCV framework
    • NumPy
    • SRGAN Model
    • CUDA Toolkit
    • Dlib

Back end

Database

    • Local File Storage

Technologies used

(Selection of the technologies can be changed when implementing based on the best approach)

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

25 of 91

System, personal, and software specification requirements

Software specification Requirement

Functional Requirement

Non-Functional Requirement

    • Low-Resolution Footage Input
    • Pre-processing of Footage
    • Visual Quality Enhancement
    • Integration with Student Monitoring System

Personal specification Requirement

    • Stakeholders – Students, Lectures (Teachers)
    • Dataset – Captured student’s videos/photos
    • Institutes – School, University

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

26 of 91

Completion of the project

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

27 of 91

What I have done in PP1

Requirement Gathering

•Finding Technologies.

•Configure development environment.

•Dataset : Image Super Resolution (from Unsplash) - https://www.kaggle.com/datasets/quadeer15sh/image-super-resolution-from-unsplash

Design Concepts

•Design Drafts.

•Design Structure.

•Follow Tutorials.

Start implementation

•Start Model Implementation

•Implent on Image Upscaling

Test the model

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

28 of 91

Train the model and implement with cuda technology

Enhance the video

    • Input the video
    • Break the video into frames
    • Enhance the frames
    • Convert the frames into video and get the enhanced video.

Test the component

Integrated with main system

    • Segment the input video and enhance the face videos.

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

What I have done in PP2

29 of 91

Implement with front end

Test the system with student segmentation

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

What I have done for final

Test the system with full system

Test component in classroom setting

30 of 91

Project Evidence

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

31 of 91

Completion of the project

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

32 of 91

Completion of the project

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Process of image upscaling output

33 of 91

Project Evidence

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

34 of 91

Completion of the project

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

35 of 91

Project Evidence

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Input

Output

36 of 91

Project Evidence

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

Input

Output

37 of 91

Project Evidence

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

38 of 91

References

[1] C. Ledig et al., “Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network,” arXiv.org, Sep. 15, 2016. https://arxiv.org/abs/1609.04802v5

[2] N. Tovar, S. (Seok-C. Kwon, and J. Jeong, “Image Upscaling with Deep Machine Learning for Energy-Efficient Data Communications,” Electronics, vol. 12, no. 3, p. 689, Jan. 2023, doi: 10.3390/electronics12030689.

[3] L. Galteri, L. Seidenari, T. Uricchio, M. Bertini, and A. Del Bimbo, “Preserving Low-Quality Video through Deep Learning,” IOP Conference Series: Materials Science and Engineering, vol. 949, no. 1, p. 012068, Nov. 2020, doi: 10.1088/1757-899x/949/1/012068.

[4] M. H. Maqsood, R. Mumtaz, I. U. Haq, U. Shafi, S. M. H. Zaidi, and M. Hafeez, “Super Resolution Generative Adversarial Network (SRGANs) for Wheat Stripe Rust Classification,” Sensors, vol. 21, no. 23, p. 7903, Nov. 2021, doi: 10.3390/s21237903.

[5] T. Sharmila and L. M. Leo, “Image upscaling based convolutional neural network for better reconstruction quality,” 2016 International Conference on Communication and Signal Processing (ICCSP), Apr. 2016, Published, doi: 10.1109/iccsp.2016.7754236.

IT20191788 | Wijesiriwardana H.G.N.D | 2023-227

39 of 91

IT20189594 Rathnayka R. K. A. R.

Specialization : Information Technology

40 of 91

Student Attendance Marking

with Face Recognition

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

41 of 91

Background

Traditional attendance tracking methods are time-consuming, unreliable, and prone to errors, leading to inaccurate attendance records.

The proposed system uses real time facial recognition technology and deep learning algorithms to accurately track and record attendance in real-time.

The system is expected to improve attendance tracking accuracy, reduce administrative workload, and provide valuable insights into student engagement and attendance patterns.

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

42 of 91

Research Problem

Traditional attendance taking methods are tedious, time-consuming and prone to human error.

Additionally, can be easily fooled by students who sign in for their absent peers, use fake identification cards.

Therefore, there is a need for a more accurate and efficient attendance tracking system that can overcome these limitations.

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

43 of 91

Novelty

Detecting and Identifyingmultiple students faces and mark attendance.

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Tracking eye gaze of each student to measure focus status

44 of 91

Specific and sub objectives

Develop a reliable face recognition algorithm to identify students in a classroom.

Identify students with their names and track individual students time.

Integrate the algorithm with a database system to efficiently store and manage attendance data.

Preprocess data before sending to ui.

Optimize the algorithm for classroom environment.

Evaluate the system's accuracy and effectiveness through real-world testing and experiments.

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

45 of 91

System Diagram

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

46 of 91

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Facial landmarks used to compare and recognize faces

47 of 91

What we have done previously in PP1

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Collect data set for facial recognition

Can detect multiple faces at a time

Implement the backend

48 of 91

Can detect and recognize multiple faces and mark attendence

Integrate component into the system

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Create a custom dataset to test the application.

Test component in real world settings eg:- classroom

What we have done previously in PP2

49 of 91

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Integrate with front end result UI

What we have done for final

Implementation of eye gaze tracker to measure focus state

Test component in classroom setting

50 of 91

IT20200206 Mallawarachchi S.M.A.

Specialization : Information Technology

51 of 91

Emotion and Behavior Detection

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

52 of 91

Background

Lectures can not focus always to emotion condition of every students.

Emotion detection technology is increasingly being used in education to support teaching and learning

By integrating an emotion recognition system into a student monitoring system, teachers can learn a lot about their students' emotional states.

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

53 of 91

Research Problem

Classroom lecturing environment is different from online lecturing environment, because teacher cannot pay attention to each student’s emotional condition and always give feedback while considering the course schedule.

How can we overcome this problem?

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

54 of 91

Novelty

Capturing emotions and head movements with multiple students.

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

55 of 91

Specific and sub objectives

Capture the emotions status of a particular student.

Develop a reliable face recognition algorithm to identify students in a classroom.

Optimize the algorithm for classroom environment.

Evaluate the system's accuracy and effectiveness through real-world testing and experiments.

Capture the head position of a student.

Preprocess the output data before send to the main component.

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

56 of 91

System Diagram

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

57 of 91

What we have done in PP1

Collected data set for students emotions

Data set: https://www.kaggle.com/datasets/msambare/fer2013

Can detect multiple emotions.

Implemented the backend

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

58 of 91

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

Preprocess output data on emotion detection - Lable the data

Implement a model to capture head possession (Student behavior)

Integrated with the main system.

What we have done in PP2

59 of 91

What we have done for final

Develop user friendly interfaces for desktop application (Front end)

Integrate front end with back end

IT20189594 | Rathnayka R. K. A. R. | TMP-23-227

Fine tune system, reduce limitations

Test in real world classroom environment

60 of 91

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

Emotion Feature Extraction Output

Output 138 features

61 of 91

References

[1] Hindawi, W. Wang, K. Xu, H. Niu, and X. Miao, “Emotion Recognition of Students Based on Facial Expressions inOnline Education Based on the Perspective of Computer Simulation,” Emotion Recognition of Students Based on Facial Expressions in Online Education Based on the Perspective of Computer Simulation, Sep. 11, 2020. https://www.hindawi.com/journals/complexity/2020/4065207/

[2] S. Fakhar et al., “Smart Classroom Monitoring Using Novel Real-Time Facial Expression Recognition System,” MDPI, Nov. 27, 2022. https://www.mdpi.com/2076-3417/12/23/12134

[3] Hindawi and Q. Yuan, “Research on Classroom Emotion Recognition Algorithm Based on Visual Emotion Classification,” Research on Classroom Emotion Recognition Algorithm Based on Visual Emotion Classification, Aug. 08, 2022. https://www.hindawi.com/journals/cin/2022/6453499/

[4] “Emotion Recognition in E-learning Systems,” Emotion Recognition in E-learning Systems | IEEE Conference Publication | IEEE Xplore. https://ieeexplore.ieee.org/document/8525872

IT20200206 | Mallawarachchi S.M.A. | TMP-23-227

62 of 91

IT20122850

Perera S.S.A

Specialization : Information Technology

63 of 91

Monitor Student's Concentration levels

IT20122850 | Perera S.S.A | TMP-23-227

64 of 91

Background

IT20122850 | Perera S.S.A | TMP-23-227

The suggested system has several benefits over conventional student monitoring methods, including the potential to give many students help and real-time feedback.

The suggested system's design, implementation, and evaluation are discussed in the study, which also highlights how well it analyzes student behavior and emotional states and produces customized feedback.

The system aims to provide reliable analysis of student behavior and emotional states, generate generalized feedback, and enhance student monitoring conditions in physical education classrooms.

65 of 91

Research Problem

How to Collect emotions and behaviors data in physical classroom?

How to perform customized separate model for each students emotions and behaviors?

How to measure a concentration level?

IT20122850 | Perera S.S.A | TMP-23-227

66 of 91

Novelty

Developing a model using emotions and behaviors to measure students concentration levels and participations.

IT20122850 | Perera S.S.A | TMP-23-227

67 of 91

Specific and sub objectives

The main objective of this research project is to create a system that uses machine learning to track students participation and concentration level in actual classroom settings.

To implement the system in a physical classroom environment.

Analyze the distractive activities of student.

IT20122850 | Perera S.S.A | TMP-23-227

68 of 91

June 2023 – Field Visit for Data Gathering

Day 01 - Field visit

Day 02- Discuss the teachers

Day 03- Filed visit

Day 04- Collecting the data

Day 05- Collecting the data

Day 06- Collecting the data

Day 07- Collecting the data

69 of 91

Data & Evaluation

IT20122850 | Perera S.S.A | TMP-23-227

70 of 91

Data & Evaluation

IT20122850 | Perera S.S.A | TMP-23-227

71 of 91

Consent form

IT20122850 | Perera S.S.A | TMP-23-227

Linlk: https://docs.google.com/forms/d/e/1FAIpQLSda8o8nxOLl_mW31XYXXaHwyfk07VxQHwKTuVBQiLaOnglfAw/viewform

72 of 91

Collected Data set

IT20122850 | Perera S.S.A | TMP-23-227

73 of 91

Data collection

IT20122850 | Perera S.S.A | TMP-23-227

74 of 91

Evidence for completion of Analytics

IT20122850 | Perera S.S.A | TMP-23-227

75 of 91

How to verify datasets

76 of 91

How to verify datasets

77 of 91

How to verify with research papers

IT20122850 | Perera S.S.A | TMP-23-227

Trabelsi, Z.; Alnajjar, F.; Parambil, M.M.A.; Gochoo, M.; Ali, L. Real-Time Attention Monitoring System for Classroom: A Deep Learning Approach for Student’s Behavior Recognition. Big Data Cogn. Comput. 2023, 7, 48. https://doi.org/10.3390/bdcc7010048

Afroze, Sadia & Hoque, Moshiul. (2020). Classification of Attentional Focus Based on Head Pose in Multi-object Scenario. 10.1007/978-3-030-33585-4_35.

Krithika L.B, ; Lakshmi Priya GG, (2016). Student Emotion Recognition System (SERS) for e-learning Improvement Based on Learner Concentration Metric. Procedia Computer Science, 85(), 767–776. doi:10.1016/j.procs.2016.05.264

78 of 91

Results

IT20122850 | Perera S.S.A | TMP-23-227

79 of 91

Results

IT20122850 | Perera S.S.A | TMP-23-227

80 of 91

Development Tools

    • PyCharm IDE for back end

    • Python
    • Tensarflow library
    • Numpy library
    • Pandas
    • CNN alogorithm

Back end

Database

    • Firebase / MongoDB

Technologies used

    • React js for front end

IT20122850 | Perera S.S.A | TMP-23-227

81 of 91

System, personal, and software specification requirements

Software specification Requirement

Functional Requirement

Non-Functional Requirement

• The system should be able to collect data from a variety of sources, such as cameras and wearable technology.

• The data must be clean and appropriate for machine learning analysis before the system should preprocess it.

• The system should be able to evaluate the data using machine learning techniques and deliver immediate feedback on the level of student interest and focus

•The system has to be simple to use and easy to use.

• The system must be secure and protect student privacy.

• There should be less downtime and the system should be dependable.

Personal specification Requirement

• Stakeholders – Students, Lectures (Teachers)

• Dataset – Captured student’s videos/photos

• Institutes – School, University

IT20122850 | Perera S.S.A | TMP-23-227

82 of 91

Model Training

IT20122850 | Perera S.S.A | TMP-23-227

83 of 91

Model Training

IT20122850 | Perera S.S.A | TMP-23-227

84 of 91

What we have done previously in PP1

Collect data set for students emotions and head movements

Research on deep learning models to train the model with an algorithm.

Research on automated measurement of affective parameters.

IT20122850 | Perera S.S.A | TMP-23-227

Implement the backend

85 of 91

Train model using local emotions, head movements data set and binary values.

Train model with csv and use binary data as emotions weight.

Test trained model using inputs from emotion and face detection components

IT20122850 | Perera S.S.A | TMP-23-227

Implement the backend

Integrate component into the system

What we have done previously in PP2

Create a full dataset values in a csv file.

86 of 91

What we have done for final

Display final result output graphically in a UI

Integrate front end with back end

Test in real world classroom environment

Generate final report output into a csv file

IT20122850 | Perera S.S.A | TMP-23-227

87 of 91

User Inteface

IT20122850 | Perera S.S.A | TMP-23-227

88 of 91

References

ICuixiang Guo, Junwu Suo, Chunguang Xu, Xinhua Yang, Liping Zhang, "Data Analysis of Physical Fitness Monitoring Based on Mathematical Models", Mathematical Problems in Engineering, vol. 2021, Article ID 8353391, 9 pages, 2021. https://doi.org/10.1155/2021/8353391 [3]

Xiaoying Shen, Chao Yuan, "A College Student Behavior Analysis and Management Method Based on Machine Learning Technology", Wireless Communications and Mobile Computing, vol. 2021, Article ID 3126347, 10 pages, 2021. https://doi.org/10.1155/2021/3126347 [1]

Ibáñez, V., Pérez, S., Silva, J., & Tamarit, S. (2019). Statistical analysis of students’ behavioral and attendance habits in engineering education. Educational Sciences: Theory and Practice, 19(4), 48 - 64. http://dx.doi.org/10.12738/estp.2019.4.004 [2]

IMaier, M.-I; Czibula, G.; One¸t-Marian, Z.-E. Towards Using Unsupervised Learning for Comparing Traditional and Synchronous Online Learning in Assessing Students’ Academic Performance. Mathematics 2021, 9, 2870. https://doi.org/10.3390/ math9222870 [4]

IYuan Zhang, Aiqiang Wang, Wenxin Hu, "Deep Learning-Based Consumer Behavior Analysis and Application Research", Wireless Communications and Mobile Computing, vol. 2022, Article ID 4268982, 7 pages, 2022. https://doi.org/10.1155/2022/4268982 [5]

Sáiz-Manzanares, M.C.; Rodríguez-Díez, J.J.; Díez-Pastor, J.F.; Rodríguez-Arribas, S.; Marticorena-Sánchez, R.; Ji, Y.P. Monitoring of Student Learning in Learning Management Systems: An Application of Educational Data Mining Techniques. Appl. Sci. 2021, 11, 2677. https://doi.org/10.3390/app11062677 [6]

IT20122850 | Perera S.S.A | TMP-23-227

89 of 91

Demonstration

90 of 91

Any Questions

91 of 91

Thank You !