CLASS LINK+ : Intelligent Online Learning Environment

CLASS LINK+ : Intelligent Online Learning Environment

Athulkrishna Suresh

Dept. of Computer Science & Engineering Toc H Institute of Science & Technology Kerala, India

Abel Andrews Aji

Dept. of Computer Science & Engineering Toc H Institute of Science & Technology Kerala, India

Anamika Ajesh

Dept. of Computer Science & Engineering Toc H Institute of Science & Technology Kerala, India

Alen Benny George

Dept. of Computer Science & Engineering Toc H Institute of Science & Technology Kerala, India

Mr. Arun Kumar

Assistant Professor, Dept. of CSE Toc H Institute of Science & Technology, Kerala, India

Abstract – Online education has become an essential part of modern learning, especially in schools and higher education institutions. While online classroom platforms provide flexibility and easy access, they face major challenges related to attendance accuracy and student participation. Most existing systems mark attendance only at login, which does not ensure that students are actively present throughout the session.

This paper introduces ClassLink+, an Android-based intel- ligent online classroom system designed to ensure accurate attendance monitoring and active student participation. The proposed system integrates real-time video communication with AI-based attendance tracking and an inbuilt smart board for note-taking. ClassLink+ continuously verifies student presence using face or person detection techniques applied to live camera feeds, reducing the need for manual monitoring by teachers.

The system architecture uses Flutter-based mobile applications for students and teachers, supported by a backend AI processing module for attendance verification. Students who remain absent from the camera view beyond a predefined threshold are auto- matically removed from the class, ensuring attendance reflects actual participation. The integrated smart board allows students to write notes during live sessions, which are saved and converted into PDF format.

Experimental evaluation demonstrates reliable attendance de- tection under varying classroom conditions. The results show that ClassLink+ improves attendance accuracy, reduces instructor workload, and enhances the effectiveness of online learning.

Index Terms – Online Education, Automated Attendance Sys- tem, Face Detection, Smart Classroom, Artificial Intelligence, Android Application

1. INTRODUCTIONOnline education has slowly become a normal part of learning in many countries, including India. Colleges, schools, and training centres now depend on online platforms to

   conduct regular classes, tests, and discussions. Students attend these classes using mobile phones, laptops, or tablets from their homes or hostels. Online learning is mainly preferred because it reduces travel time, saves money, and allows flexible learning.

   Although online classes are convenient, they also bring several challenges. In a physical classroom, teachers can easily see whether students are present and paying attention. In online classes, this becomes difficult because teachers only see screens and login names. Many students join the class and stay logged in even when they are not actually attending. Because of this, the quality of learning reduces and teachers find it hard to manage discipline.

   Another issue in online education is the lack of proper tools to support learning. Students usually depend on separate applications or notebooks to take notes during classes. Switch- ing between multiple apps often breaks concentration. These problems show that online education needs better systems that can handle attendance properly and also support students during learning.

   This project is focused on developing an improved online classroom system that helps teachers manage attendance more effectively and helps students stay focused during online classes.

   1. Problem StatementMost existing online classroom systems follow a very basic method for attendance. Attendance is usually marked when a student joins the class. After joining, the system does not check whether the student remains present or active. A student

      can leave the device, attend other activities, and still be marked as present. This creates false attendance records and reduces the seriousness of online education.

      Teachers also face difficulties in monitoring students manu- ally. In classes with many students, it is almost impossible to track who is paying attention and who is not. Manual checking consumes time and still does not give accurate results. Another common issue is note-taking. Students often use different ap- plications to write notes, which causes distraction and affects understanding during live sessions.

      Because of these limitations, current online classroom systems fail to provide reliable attendance data and proper learning support. This creates a gap between online learning and traditional classroom learning.

      The major problems identified in existing systems are listed below:

      • False or Proxy Attendance: Students can remain logged in without actually attending the class.
      • No Presence Verification After Login: The system does not check whether students stay present till the end.
      • Manual Monitoring Difficulty: Teachers cannot effec- tively observe students in large online classes.
      • Low Student Attention: Students may stay connected but not mentally involved in learning.
      • Distraction Due to External Note-Taking: Switching between apps reduces focus.
      • Unreliable Attendance Records: Attendance data does not reflect real participation.
   2. Proposed SolutionTo overcome the problems faced by existing systems, the proposed solution ClassLink+ is designed as a smart online classroom application. The main goal of the system is to ensure that attendance is based on actual student presence and not just login details.

      The system uses live video input to observe student presence during the class. Instead of checking attendance only once, the system continues to monitor whether the student remains in front of the device. If a student is not detected for a certain amount of time, the system automatically removes the student from the class. This encourages students to stay attentive and reduces misuse of the platform.

      Along with attendance monitoring, ClassLink+ also pro- vides learning support features. The system includes an inbuilt writing board that allows students to write notes during the class itself. This reduces the need to use external applications. The notes written during the class are saved automatically and converted into PDF files so that students can use them later for revision.

      Attendance details are stored securely and attendance re- ports are generated automatically at the end of each class. This reduces manual work for teachers and helps maintain accurate records.

      The important features of the proposed system are listed below:

      • Video-Based Presence Checking: Student presence is verified using live video instead of login details.
      • Continuous Monitoring: Presence is checked through- out the class duration.
      • Automatic Removal of Inactive Students: Students who are not detected are removed automatically.
      • Integrated Writing Board: Students can take notes within the same application.
      • Automatic PDF Note Saving: Notes are saved and converted into PDF format.
      • Automatic Attendance Report Generation: Attendance reports are created without manual effort.
2. LITERATURE REVIEWThe rapid growth of online education has increased the need for intelligent systems that support effective learning, moni- toring, and engagement in virtual classrooms. Recent studies highlight the role of smart technologies, cloud platforms, and high-speed communication such as 5G in transforming modern teaching methodologies [10], [11], [13]. Although these ad- vancements improve accessibility and scalability, most online platforms still rely on login-based attendance mechanisms, which fail to ensure continuous student participation.
   1. Online Learning and Smart Education ModelsSmart education frameworks integrate digital technologies such as mobile applications, cloud computing, artificial in- telligence, and IoT to enhance learning experiences [2], [3], [11]. These models focus on personalized learning, real- time interaction, and automated academic processes. However, many smart education systems emphasize infrastructure and content delivery rather than continuous attendance verification, limiting their effectiveness in online classroom environments [12], [13].
   2. AI-Based Attendance Monitoring SystemsSeveral research works propose automated attendance sys- tems using face recognition and computer vision techniques. Methods such as Haar Cascade, Local Binary Pattern (LBP), FaceNet, and deep learning-based approaches have been widely used to identify students using camera feeds [4], [6][8]. Group-based face recognition and tracking methods further improve robustness by handling movement and occlu- sion during learning sessions [1]. While these systems reduce proxy attendance and manual effort, many perform attendance verification only at the beginning of the session and lack continuous monitoring.
   3. Student Engagement and Focus MonitoringBeyond attendance, researchers have explored monitoring student focus and engagement during online lectures using AI- based techniques. Facial expression analysis, eye movement tracking, and emotion classification using CNN and MobileNet models have been studied to assess student attentiveness [5], [9]. Although these approaches provide valuable insights, they often require high computational resources and raise privacy

      concerns, making real-time deployment challenging in large- scale online classrooms.

   4. Limitations of Existing Systems and Research GapsFrom the reviewed literature, it is evident that most existing systems address attendance monitoring, engagement analysis, or online content delivery as separate components [1], [6], [10]. Continuous presence verification is often missing, allow- ing students to remain marked present despite being inactive. Additionally, learning support tools such as note-taking are typically handled through external applications, which can distract students and reduce engagement during live sessions.
   5. Motivation for the Proposed ClassLink+ System

   The identified gaps highlight the need for an integrated online classroom platform that combines live video classes, continuous AI-based attendance monitoring, and learning sup- port features within a single system. Motivated by the lim- itations of existing approaches [3], [6], [13], the proposed ClassLink+ system aims to provide an Android-based solution with real-time attendance verification, automatic removal of inactive students, and an inbuilt smart board for note-taking. This integration offers a more reliable and practical approach to modern online education.

3. SYSTEM DESIGN AND METHODOLOGYThis section describes the overall design of the proposed system ClassLink+ and explains the methodology followed for its implementation. The system is designed to support online classes with automated attendance monitoring and interactive learning features.
   1. System DesignThe system design explains how different components of ClassLink+ are structured and how they interact with each other. The system follows a clientserver architecture, where the client side consists of applications used by students and teachers, and the server side handles video processing, atten- dance monitoring, data storage, and report generation.
      1. High-Level Architecture: The high-level architecture of the ClassLink+ system consists of student and teacher appli- cations connected through an online classroom environment. Live video communication is established between users during the class. Student video streams are processed by the AI-based attendance module to verify continuous presence. Attendance data and reports are stored securely in the database. The overall structure of the ClassLink+ system is illustrated in Fig. 1, which shows the interaction between student and teacher applications with the server-side modules.
      2. Module-Wise System Design: The ClassLink+ system is divided into multiple modules to simplify development and improve efficiency.
      3. Student Module: The student module allows students to log in, join online classes, stream video, take notes using the smart board, and view their attendance status.Fig. 1: High-Level Architecture
      4. Teacher Module: The teacher module enables teachers to start and manage classes, monitor students, and access attendance reports after the class ends.
      5. Video Streaming Module: The video streaming module is responsible for enabling live interaction between students and teachers during online classes. It supports real-time video transmission with minimal delay to ensure smooth communi- cation. Student video feeds are sent to the server side, where they are further utilized for attendance verification without disturbing the ongoing class session.
      6. Attendance Monitoring Module: This module continu- ously monitors student participation during the class using computer vision techniques. By analyzing live video streams, the system checks for the presence of students at regular intervals. Attendance is recorded only for students and not for the teacher, ensuring fairness and accuracy in attendance tracking.
      7. Smart Board Module: The smart board module provides an interactive space for students to write notes and illustrate concepts during live sessions. This feature helps improve engagement and allows students to actively participate in the learning process. All written content is saved automatically and can be accessed later for revision and reference.
      8. Database and Report Module: The database and report module is used to securely store attendance records and session-related data generated during each class. It ensures that all information is properly organized and maintained for future use. This stored data forms the basis for generating attendance summaries and reports.
      9. Report Generation Module: This module converts stored attendance data into well-structured PDF reports. These reports provide teachers with clear insights into student attendance andparticipation throughout the class. The generated reports can be easily accessed, downloaded, and used for academic record keeping.
   2. MethodologyThe methodology describes the overall working of the proposed system.
      1. System Flow:The operational workflow of the system is further explained using a flowchart, as shown in Fig. 2, detailing the sequence of actions performed during a live class session.Fig. 2: System flowchart

         The system flow starts with user login and class joining. Once the class begins, live video streaming is activated. Student presence is continuously monitored during the session, while the teacher conducts the class normally. Attendance status is updated dynamically based on student activity.

      2. Attendance Detection Process: The attendance detection process is based on continuous monitoring of student presence using live video frames. Video frames are extracted at regular intervals and analyzed to detect the presence of a face or per- son. If the student is detected, attendance monitoring continues without interruption.
      3. Attendance Calculation: Attendance percentage is cal- culated using the active participation time of the student.Let:
         • Ta be the active time of the student
         • Tt be the total duration of the classThe attendance percentage is calculated as:

           Attendance(%) = Ta × 100 (1)

           Tt

      4. Removal Condition: If a student is not detected for a fixed threshold time, the system removes the student from the class to prevent false attendance.If Absent Time > Threshold Remove Student (2)
      5. Smart Board and Note Generation: During the class, students can write notes using the smart board feature provided in the application. The notes are saved automatically in real time and converted into PDF format after the class ends for future reference.
      6. Attendance Report Generation: After the class ends, attendance data is collected and processed automatically. Attendance reports are generated in PDF format and made available to teachers for review and record keeping.

   This methodology ensures accurate attendance tracking, reduces manual effort, and improves the overall effectiveness of online classes.

4. Implementation and Experimental ResultsThis section describes the implementation of the ClassLink+ attendance tracking system and presents the experimental re- sults obtained during testing. The focus of the implementation is on accurate student attendance detection using face or person detection in an online classroom environment.
   1. Data Collection and PreprocessingA custom dataset was created for the ClassLink+ attendance detection module. The dataset consists of student face images captured during online classes using Android device cameras. Images were collected under different conditions such as normal lighting, low lighting, face movement, and partial occlusion to ensure robustness.

      Approximately 300 images were collected per student, resulting in a dataset of around 1500 images. All images were resized to a fixed resolution and normalized before training.

      TABLE I: Dataset Details for Attendance Detection

      Parameter	Description
      Dataset Type	Custom student face dataset
      Number of Students	5
      Images per Student	Approximately 300
      Total Images	Approximately 1500
      Image Conditions	Normal light, low light, different angles
      Image Source	Live camera capture

      1. Data Augmentation: To prevent overfitting and improve generalization, several data augmentation techniques were applied during training.TABLE II: Data Augmentation Techniques
         

         Technique	urpose
         Horizontal Flip	Handles left and right face orientation
         Brightness Adjustment	Simulates lighting variations
         Contrast Adjustment	Handles camera quality differences
         Random Crop	Handles distance variations
         Resize	Maintains input consistency

   2. Training ConfigurationThe attendance detection model was trained using a deep learning framework on a GPU-enabled system. The training parameters were carefully selected to balance accuracy and computational efficiency. The trained model is deployed on the backend server, while the Android application runs on standard mobile devices. Heavy computation is handled by the backend to ensure smooth performance on student devices.
      1. Application Interface:Fig. 3: Teacher Dashboard Interface of the ClassLink+ Appli- cation

         Figure 3 shows the teacher dashboard of the ClassLink+ application. It provides a quick overview of active classes and registered students, allowing instructors to access key features and monitor classroom activity from a single interface.

         Fig. 4: Registered Students List with Real-Time Activity Status

         Figure 4 presents the registered students module of the ClassLink+ system. This screen displays student details along with real-time activity status, helping instructors track partic- ipation.

         Fig. 5: Digital whiteboard

         Figure 5 shows a sample handwritten input created using the digital whiteboard feature of the ClassLink+ system. This module allows students to write or draw in real time during online classes, supporting interactive learning and note sharing.

   3. Qualitative Model EvaluationQualitative evaluation was performed by testing the system during live online classroom sessions.

      Fig. 6: Live online classroom with multiple students

      Figure 6 demonstrates the live online classroom environ- ment with multiple students. The system continuously moni- tors student presence and updates attendance status based on face visibility during the session.

      The system was observed under different conditions such as student movement, temporary face absence, and lighting changes. The model successfully detected student presence when the face was visible and increased the absence timer when the face was not detected.

      Fig. 7: Successful Face Detection

      Figure 7 illustrates a live classroom scenario where the system successfully detects a students face. When the face is visible, the student is marked as active and attendance tracking continues without interruption.

      Fig. 8: Absence Detection

      Figure 8 shows the absence detection mechanism of the system. When no face is detected, a countdown timer is triggered, warning the student before automatic removal from the session.

   4. Performance MetricsThe following metrics were used to evaluate the perfor- mance of the attendance detection system.

      TABLE III: Performance Metrics

      Metric	Description
      Detection Accuracy	Correct detection of student presence
      False Absence Rate	Incorrect absence marking
      Response Time	Time taken to detect presence
      Attendance Reliability	Consistency over class duration
      System Stability	Performance during long sessions

   5. Attendance Detection ResultsThe trained model achieved high accuracy in detecting student presence under normal classroom conditions. Minor errors were observed during extreme lighting changes and heavy face occlusion.

      TABLE IV: Attendnce Detection Accuracy

      Condition	Detection Accuracy
      Normal Lighting	94%
      Low Lighting	89%
      Face Movement	91%
      Partial Occlusion	87%
      Average Accuracy	90%

   6. System Response Analysis

   The system response to different student behaviors during online classes is summarized below.

   TABLE V: System Response to Student Activity

   Scenario	System Response
   Face detected	Student marked active
   Temporary absence	Absence timer increased
   Prolonged absence	Student removed from class
   Class completion	Attendance report generated

   Overall, the experimental results confirm that the ClassLink+ system provides reliable and accurate attendance tracking for online classroom environments.

5. Conclusion and Future Scope

The ClassLink+ project successfully demonstrates a func- tional prototype of an intelligent online classroom system. By combining real-time video communication with AI-based attendance monitoring and an inbuilt smart board, the system addresses common issues found in existing online learning platforms. Unlike traditional systems that depend only on login-based attendance, ClassLink+ verifies continuous student presence using face or person detection, which helps reduce proxy attendance and inactive participation.

The system removes the need for manual attendance check- ing by teachers and generates attendance reports automatically at the end of each session. The integrated smart board allows students to take notes during live classes, and these notes are saved and converted into PDF format for future use. Overall, the project shows that artificial intelligence can be effectively applied to improve attendance accuracy, discipline, and learning quality in online education without the need for additional hardware.

Although the current implementation of ClassLink+ achieves its primary objectives, several enhancements can be considered for future development.

• Improved Face Detection Accuracy: Enhancing detec- tion performance under low lighting conditions, different camera angles, and unstable network environments.
• Multi-Face and Group Attendance Handling: Extend- ing the system to support attendance tracking for multiple students using shared devices or group learning setups.
• Edge AI Optimization: Optimizing AI models using techniques such as quantization or lightweight inference to enable on-device processing and reduce server depen- dency.
• Emotion and Attention Analysis: Adding basic emo- tion and attention detection to better understand student engagement during online classes.
• Class Recording and Playback: Integrating class recording and cloud storage so students can revisit recorded sessions along with saved notes.
• Cross-Platform Support: Expanding the system to sup- port iOS and web-based platforms for wider adoption.

REFERENCES

1. N. Bao, X. Wei, Y. Ma, and Z. Liang, Online group-based face recognition and tracking algorithm using Kalman filtering in team learning mode, in Proc. Int. Conf. Image Processing, Computer Vision and Machine Learning (ICICML), 2024, pp. 10781085.
2. K. Zala, S. Kothari, H. Patel, A. Bhola, and B. Acharya, Smart education model: Future learning and teaching, in Proc. IEEE 11th Region 10 Humanitarian Technology Conference (R10-HTC), 2023,pp. 6469.
3. A. Dehbi, A. Bakhouyi, R. Dehbi, and M. Talea, Towards a smart learning environment: A model for smart education using IoT and AI, in Proc. IEEE Int. Symp. Advanced Electrical and Communication Technologies (ISAECT), 2024.
4. R. Mohamed, J. U. Jafni, and S. N. M. Rum, Real-time face recognition system in smart classroom using Haar Cascade and Local Binary Pattern model, in Proc. IEEE Int. Conf. Advanced Creative Networks and Intelligent Systems (ICACNIS), 2022.
5. S. AbuAisheh, Y. W. Teh, and L. Shuib, Monitoring university stu- dents focus during online lectures using artificial intelligence tools, in Proc. IEEE Int. Conf. Computational Intelligence Approaches and Applications (ICCIAA), 2025.
6. D. Vengaimarbhan, C. G. Krishnaa, I. Nandhini, S. Dhenathayalan, andS. raiarasu, An efficient online attendance monitoring system using FaceNet-based facial recognition technology, in Proc. IEEE Int. Conf. Artificial Intelligence and Machine Learning Applications (AIMLA), 2025.
7. S. Patil, A. Jani, D. Jondhale, D. Pathak, and A. S. Savyanavar, Stu- dents attention and alertness monitoring system using AI, in Proc. 2024 MIT Art, Design and Technology School of Computing International Conference (MITADTSoCiCon), Pune, India, Apr. 2024,pp. 16.
8. A. K. Sirivarshitha, K. S. Priya, K. Sravani, and V. Bhavani, An approach for face detection and face recognition using OpenCV and face recognition libraries in Python, in Proc. 9th Int. Conf. Advanced Computing and Communication Systems (ICACCS), Coimbatore, India, 2023, pp. 12741279, doi: 10.1109/ICACCS57279.2023.10113066.
9. R. Yadav B., S. Eliyas, and S. K. M., Classifying emotions and engagement in online learning based on single facial recognition using CNN and MobileNet-v2, in Proc. OPJU Int. Technology Conf. on Smart Computing for Innovation and Advancement in Industry 4.0 (OTCON), 2024, pp. 17, doi: 10.1109/OTCON60325.2024.10687723.
10. R. Yan, Application and analysis of education and teaching mode based on 5G and smart technology, Scientific Programming, vol. 2022, pp. 1 10, 2022.
11. I. Lyapina, E. Sotnikova, O. Lebedeva, T. Makarova, and N. Skvortsova, Smart technologies: Perspectives of usage in higher education, Interna- tional Journal of Educational Management, vol. 33, no. 3, pp. 454461, 2019.
12. L. Feng, The analysis of computer teaching mode based on modern educational technology, in Proc. Int. Conf. on Educational and Infor- mation Technology, 2018.
13. K. Demir, Smart education framework, Smart Learning Environments, vol. 8, no. 1, pp. 136, 2021.
14. T. Li, X. Liu, and P. Zhang, Deep learning-based real-time face detection and recognition system for smart attendance, IEEE Access, vol. 8, pp. 2193021941, 2020.
15. M. Abdurrahman, T. S. Gunawan, M. Kartiwi, and M. A. Rahman, Development of IoT-based face recognition attendance system using FaceNet, Indonesian Journal of Electrical Engineering and Computer Science, vol. 23, no. 3, pp. 12501260, 2021.
16. A. Kumar and R. Yadav, Facial recognition attendance system using MTCNN and FaceNet model, in Proc. Procedia Computer Science, vol. 180, pp. 168175, 2021.
17. S. M. Lajevardi and H. R. Wu, Facial expression recognition in perceptual color space, IEEE Transactions on Image Processing, vol. 21, no. 8, pp. 37213733, 2012.
18. G. Yang and T. S. Huang, Human face detection in a complex background, Pattern Recognition, vol. 27, no. 1, pp. 5363, 1994.
19. M. H. Yang, N. Ahuja, and D. Kriegman, Face detection using mixtures of linear subspaces, in Proc. Fourth IEEE Int. Conf. on Automatic Face and Gesture Recognition, pp. 7076, 2000.