Smart Proctoring System for Secure and Fair Online Exams

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

Smart Proctoring System for Secure and Fair Online Exams

Shreeram Mutukundu1, Ankush Tiwari2, Adarsh Kumbhar3 , Varsha Kulkarni4

1Student, Dept. of Computer Engineering, JSPM’s Imperial College of Engineering and Research, Wagholi, Pune, Maharashtra, India

2Student, Dept. of Computer Engineering, JSPM’s Imperial College of Engineering and Research, Wagholi, Pune, Maharashtra, India

3Student, Dept. of Computer Engineering, JSPM’s Imperial College of Engineering and Research, Wagholi, Pune, Maharashtra, India

4Professor, Dept. of Computer Engineering, JSPM’s Imperial College of Engineering and Research, Wagholi, Pune, Maharashtra, India

Abstract - In recent years, online examinations have become more and more popular because of their flexibility. Especially under the effect of COVID-19 pandemic, guaranteeing exam cheat-free becomes a big challenge for educational institutions. In this paper, we propose the AIbased proctoring that does not require continuous human being supervision. The system uses neural networks and machine learning to detect inevitable actions in a test such as eye movement tracking, mouth movement tracking and device using checking etc. Experimental results show that our developed system decreases cheating rate and outperforms other human-based proctoring approaches.

Key Words: AI proctoring, cheating detection, machine learning, online exam security, proctoring automation, real-timemonitoring

1. INTRODUCTION

This research targets with a growing challenge of academic dishonest that has become evident in remote exams, especially after the expansion of online education. Thatis,thetraditionalmethodsforexamproctoringbased on human invigilators are no longer applicable and feasibleinthedigitalera.Henceweproposeaninnovative solution, namely an AI-empowered proctor system as an integrated part in online exam environments to monitor studentsandmaintaintheexamintegrityduringreal-time. Byemploying advancedalgorithms and machinelearning, AI proctors can recognize facial expressions indications, distinguish screen activities as well as capturing backgroundvariancesevenmoresufficientlythanhuman.

ThecruxofourprojectliesinmergingAIwithMERNstack to get a scalable and secure structure for creating an efficient real-time exam monitoring system. The AI track, evaluates and constantly monitor’s the examinee throughout the examination duration. Whenever AI finds anydeviationsormalpracticeslikeunauthorizedoldingof objects, usage of foreign material or accessing external devices it will alert instantly which together helps us

createacheatingfreeenvironmentalongwithmaintaining excellent user experience. This combination, mixture of artificial intelligence by keeping the web technologies suitable for all kinds of examinations as there are numeroustypesoftestsconductednowadays.

By converting from a reactive monitoring to a proactive surveillance, our proposed research demonstrates a new methodologyofonlineexamsecuritywheretheAIproctor doesn'tonlywatchtheexaminationbutinteractswiththe examinationandreportsaboutanysuspicionatfirsthand. Such advantage allows educational institutes to conduct secure exams for any number of examinees in different campuses or even time zones. This will be applicable as long as online based-learning becomes more prevalent over conventional courses. And also for e-assessment aspects this manuscript will have an opportunity to establish competition among future works which target ensuringfairness,accessibilityandtrustworthiness.

2. LITERATURE

In paper [1], authors addressed problem of online education, which became especially popular during COVID-19 pandemic all educational organizations were forced to change teaching process into distant mode and problems associated with academic integrity went up. As for students it now became easier to cheat the system usingdifferentserviceswhichcanwritestudent’spaperor makeexam insteadofhim, developersshouldurgentlydo something in order to use AI solutions, namely proctorlike systems during test administration on line and especially remotely over the Internet. For that kind of proctoring because of safety requirements new nonintrusive techniques are needed like face recognition and behaviouranalysistopreventfromcheating.

In [2], the authors proposed an AI-based online exam proctoring system to guarantee the credibility and securityofaremoteexam.Tomonitorthatstudentsdonot cheatduringtheexam,awebcamandmicrophone,aswell as sophisticated algorithms, are used. The authors

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

introduced a concept called “computer isolation” where studentscannotopenorswitchtabs/anyotherapplication while they take the exam. For this purpose, they propose AEPS system (AI- based Exam Proctoring System) which usesartificialintelligentbaseddown-streamedprocessing on facial expression recognition, head movement of candidates and object detection to detect any suspicious activities carried out by candidate during an examination ensuringitsreliability.

Research paper [3] describes the usage of live proctoring where trained supervisors view and listen to students in real-time through audio/video feeds during online exams to detect any suspicious behaviors, such as presence of unauthorized devices or unusual location changes. Another method called recorded proctoring is also explored here, in which students can take the exam at their convenience and proctors review their sessions afterwards for any suspicious activities.Regular or recorded proctoring are good way to prevent impersonation or cheating but it also adds additional burden on post-exam human resources and increases the cost.

In paper [4], the authors present the challenges of designing and developing AI based proctoring systems. They use biometric technologies such as fingerprint recognition or behavioural biometrics to perform authenticationandverifytheidentityofcandidatestostop anykindofimpersonation.Gesturerecognitionandobject detectionarealsobeingcombinedwithAItosupervisethe activities of students while they are taking exams so that wecangetmoreaccurateresultswithoutanyhassle.

The research in [5] also discusses the application of AI in proctoring systems. It explains that machine learning algorithms can be used to detect fraud by monitoring students’ behaviours; for example, typing cadence and facialexpressionscanindicatewhetherastudentistrying tocheat.Thepaperalsonotesthatreal-timedetectionand prevention of cheating requires a multi-layered monitoring approach utilizing audio and video analytics, aswell.

Instudy[6],authorspresentedahybridproctoringmodel which uses both AI and human. In this model, real time supervised interfaces (systems like “ProctorU”) are providedtotheliveproctorsforcontinuousmonitoringof the test takers along with the suspicious activities detection algorithms based on AI. This hybrid model ensures high accuracy and trust level of proctoring by maintaining flexibility provided by AI-based system and securityensuredviahumaninvolvement.

Paper [7] presents another AI-based proctoring solution, the EU-funded TESLA project. It uses multiple biometrics like voice recognition, eye movement tracking and typing patterns for candidate authentication. These methods can

also be used as part of the proctoring process to validate the identity of the candidates and the authenticity of answers given by them during an examination. There is a strongemphasisonsecurityinonlineassessmentbutlittle work has been done so far to come up with foolproof systems.

Lastly, the use of advanced object detection system i.e. YOLO (You Only Look Once) to monitor the exam room and identify presence of suspicious objects that might constitutecheatingduring examinationsconductedonline arediscussedin[8].ThesuitabilityofusingAIresearchby object detections system using YOLO is highlighted particularly when incorporated with AI algorithms providing real time inference. This paper indicated the potential inclusion of AI surveillance system as a major component to enforce academic honesty for online education as it is foreseen there will be increased in demandforonlinelearninginfuture

3. PROBLEM STATEMENT

Thefastdevelopmentofonlineeducation,especiallyunder the influence of COVID-19, raises a new challenge –protecting academic integrity on virtual exams. More convenience in learning process brings also much higher threat of dishonesty when students apply diverse, covert practices to defeat the exam safeguards. Existing proctoring solutions are beneficial but not designed for such multifaceted cheating scenarios like these presented today. We need more advanced and scalable solution which would be provided by Artificial Intelligence technology that will guarantee monitoring and securing online exams on massive scale. The aim is to develop AI driven proctoring system that will not only detect if examinee is cheating but also ensure fair and equal conditions for all examinees thus keeping examinations reliableindigitaltransformationera.

The increasing reliance on online examinations has laid bare some inherent vulnerabilities of examination administration, putting academic integrity at serious risk. Conventionalproctoringmechanismsareeithermanualor semi-automated and are not well attuned to the massive scale and complicated nature of virtual test-taking, therebyenablingvariousformsofmalpractices.Moreover, theeffectivenessofhumaninvigilatorsisbeingquestioned increasingly: fatigue, inattention/oversight and varying skill-sets would result in inconsistent monitoring. Concerns related to invasion of privacy have also exacerbated tensions around students' acceptance of any invasive technology that invades their personal space during an examination process. In view of challenges aforesaid,thereisanurgentneedforanadvanced,robust, and non-intrusive solution that can ensure fairness and securityinonlineexaminations.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

An AI-enabled proctoring system appears as a viable option; however, it needs to be judiciously designed so as maintain balance among security efficacy assurance(s), individualprivacypreservation,andusertrustbuilding.

4. STUDIES AND FINDINGS

A. Study on the Effectiveness in Higher Education:

This study was carried out by the researchers from the Indian Institute of Technology (IIT) to investigate how well AI proctoring fared during remote examinations for engineering and management students. It was found that academicians perceived that AI reduced cheating by 70% whencomparedtoregularonlineexamsheldwithoutany kindofproctoring.TheperceptionwasthattheuseofAI’s facial recognition and behaviour tracking capabilities helped to uncover suspicious behaviours, for example, unusual eye movements and instances where a student switched betweenvariousscreens.Studentsrevealedthat AI’s objectivity had made them feel that such exams were more fair but poor internet connection was an issue highlightedbybothacade-miciansandstudentsingeneral.

B. Findings on the User Experience:

A study conducted by University of Delhi measured user experience of students and faculty with AI proctoring usage.500 students and 100 faculty users provided their feedback. Students reported an increased sense of anxietyin AI proctored examination, expressing intrusive monitoring due to cameras and sensors. However, faculty had a favorable opinion related to features where it automated attendance and suspicious behavior alerts; 85% percent of the teachers believed that use of AI will not compromise academic integrity while conducting examinations, but similar perception among only 60% students. Fig.4.1:BoxPlot(survey)

This study reviewed survey research to determine how common it is for university students to admit cheating in online exams, and how and why they do it. We also assessed whether these self-reports of cheating increased during the COVID-19 pandemic, along with an evaluation of the quality of the research evidence which addressed these questions. 25 samples were identified from 19 Studies, including 4672 participants, going back to 2012. Online exam cheating was self-reported by a substantial minority (44.7%) of students in total. Pre-COVID this was 29.9%, but during COVID cheating jumped to 54.7%, althoughthesesamplesweremoreheterogenous.

5. METHODOLOGY

It has a workflow which is structured and well defined becauseitisnecessarytotakecareofexam’sintegrityand securityinvirtualenvironmentalsouser’sprivacy.Sohere given method will give you step by step details of approachfollowedforimplementationoftheproject:

The system requires following Hardware and Software components.

5.1 SOFTWARE REQUIREMENT

i. VS-Code(IDE)

ii. TechStack:

a. Frontend:Reactjs,Redux-toolkit,Material UI,JavaScript.

b. Backend:Nodejs,Express.js

c. Database:MongoDB

d. AI/ML Model:TensorFlow.js,COCO-SSD

e. Deployment:Vercel

f. File Handling:Multerforuploads

g. Authentication:JWT

5.2 HARDWARE REQUIREMENT

i. Processor:-Intel5ᵗʰGenerationorabove ii. Memory:-8GBorabove iii. HardDisk:-100gb

Fig.5.1:DataFlowDiagram

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

A. User Authentication and Authorization:

 Step 1: Students have to login before starting the examwithsecurecredentials.

 Step 2: Multi-factor Authentication (MFA) based Identityverification-Useyourbiometric(faceMe Photoorfingerprint),EnteruniqueID.

 Use of email or mobile-based OTP (One-Time Password).Thisway,youaregrantedwithaccess to exam platform and your session is locked for anyunauthenticatedaccessfurther.

B. Environment and System Check:

 Step1:Thesystemwillscantheuserenvironment through the webcam in order to detect multiple monitors, mobile phones, or any unauthorized aidswhichcouldbeusedforcheatingpurposes.

 A browser lock is initiated which will not allow usertoopennewtab,useanyshortcutkeysoruse external software’s etc. System keeps a check on CPU usage if in case any background application helpingindoingmalpractice.

 An initial network test happen to make sure that they have good connectivity and there won’t be anydisturbanceduringtheentireproctoring.

C. Real-Time Monitoring with AI Algorithms:

 Step1:Oursystemwillkeeponmonitoringuser’s activity during examination via webcam, microphone and screen sharing. Our AI algorithms will be continuously monitoring your facial expressions, eye movement, body gestures etc. it can detect any kind of suspicious activity likelookingfrequentlyhereandthereorspeaking something.

 Step 2. Audio detection keeps check if any abnormal sound or voice clue e.g. external help, conversationandotherdistractingvoicesetc.

 Step3.RecordingofscreenactivityandAIdetects if any unusual pattern occurred like copy paste text,tabswitchoraccessingrestrictedcontent.

D. Object Detection Model (COCO-SSD)

 Step1:TheCOCO-SSDmodelisintegratedintothe system to perform real-time object detection during the examination. It continuously scans the exam environment to identify unauthorized objects such as mobile phones, extra faces, or externaldevices.

 Step 2: The AI model processes webcam feeds frame by frame, ensuring low latency and high accuracyinobjectrecognition.Detectedviolations areloggedinstantlyforreview.

 Step 3: Performance evaluation is conducted using metrics like accuracy, precision-recall, and F1-score to maintain an optimal balance between detectionefficiencyandfalsepositives.

E. Automated Screenshot Capture and Logging







Step 1: The system captures periodic screenshots of the examinee’s screen and webcam feed to maintainavisualrecordthroughouttheexam.

Step 2: Each screenshot is encoded into Base64 format,optimizingstoragewithoutcompromising imageclarity.

Step 3: Using Multer, the system securely stores imagesontheserverwhileloggingmetadatasuch as timestamps, detected anomalies, and session detailsinMongoDB.

F. Violation Detection and Handling







Step 1: The AI-powered monitoring system continuously evaluates user behavior, defining strict thresholds for detecting exam rule violations.

Step 2: If an examinee’s face is absent for an extended period, multiple faces appear, or a mobile phone is detected, the system flags it as a potentialviolation.

Step 3: Upon reaching predefined violation thresholds,thesystemcanautomaticallyalertthe invigilator, issue a warning, or terminate the exam,dependingontheseverityofthebreach.

G. Code Execution







Step 1: The platform enables real-time execution of programming languages, including Python, JavaScript, and Java, within a secure coding environment.

Step 2: The system instantly identifies and highlights syntax errors, allowing students to make necessary corrections without external assistance.

Step 3: In the case of runtime errors, the system logs error details and provides structured feedback to ensure transparency in execution while preventing any unauthorized debugging tools.

H. Backend API Design







Step 1: The system includes API endpoints for logging and retrieving cheating incidents, ensuringseamlessintegrationwiththeproctoring platform.

Step 2: The POST /api/exams/cheatingLogs endpoint is used to store detected violations, includingscreenshotsandmetadata.

Step3:The GET /api/exams/cheatingLogs/:examId endpoint retrieves all logged violations for a specific exam, while POST /api/exams/logViolation updates the violationcountperexaminee.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

I. Suspicion Detection and Alerts:

 Step 1: Behaviors are evaluated by AI driven algorithms with comparison to pre-defined thresholdsofsuspicion.

 Step 2: In case of any irregularities (like prolonged eye deviation, or multiple faces detected in the frame), automated alerts are sent by the system to the invigilators for further review.

 Step 3: Invigilator is provided with a live dashboard, where he can view each student in detail and high-risk candidates are flagged automatically so that immediate action can be taken.

J. Intervention and Logging:

 Step 1: In the case of repeated suspicious activities the invigilator is prompted to intervene by either pausing the exam or communicating directlywithstudentthroughtheplatform.

 Step 2: All suspicious incidents are automatically recorded with timestamps and evidence (videos, screenshots,audio)forpost-examreview.

K. Post Exam Analysis and Report Generation:

 Step 1: Once the exam is over, AI models will process the data and generate a detailed report which contains the metrics on student behavior, flagged incidents, system performance and networkstabilityduringtheexam.

 Step 2: Flagged incidents are reviewed by proctorsandifanyviolationsaredetectedduring the exam, students are notified. Step 3: The report is safely kept with audit privacy controls so that authorized only people haveaccesstoit.

L. Data Encryption and Privacy Measures:

 Step1:Allcollecteddata(videorecordings,audio, screenshots) get encrypted and stored in a secured manner to comply with privacy laws and institutionaldatapolicies.

 Step 2: We have tried to intrude into Privacy as minimum as we can. We use minimal data to configure exam monitor and do not capture or dealwithanystudent’spersonaldata.

 Step3:Studentsknowwhatisbeingcapturedand will have access to their monitoring results for a duration after exams again complete transparency&trustrightthere.

M. System Feedback and Continuous Learning:

 Step1:TheAImodel isupdatedafterevery exam session, by machine learning algorithms trained on detected cheating incidents as well as user patterns, in order to improve future proctoring sessions.



Step2:Feedbackfromstudentsandinvigilatorsis gatheredinorderto improvetheuserexperience and to fine-tune AI based suspicion detection sensibilities.

6. PROTOTYPE DESIGN

Fig.6.1:ExamProctoredSystemPrototype

7. FEATURES

1. Real-time Webcam Monitoring – The system captures a continuous live video stream from the user's webcam, ensuring that the examinee remains present throughout the exam. This helps maintain integrity by detecting any unusualmovementsorabsences.

2. COCO-SSD Object Detection – Using COCO-SSD, the system actively scans for unauthorized objects such as mobilephonesoradditionalfaces.Thisreal-timedetection enhancessecuritybypreventingcheatingattempts.

3. Automated Screenshot Capture – The system takes periodic screenshots whenever suspicious activity is detected, providing visual proof of any potential rule violations.Theseimagesaresecurelystoredforreviewby examiners.

4. Violation Detection and Logging – Any detected violations,suchasmultiplefaces,absenceoftheexaminee, or mobile phone usage, are logged in real-time. This ensures a structured record of infractions for post-exam analysis.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

Fig.7.2:Mobilephonedetectionwhileexamisongoing

Fig.7.3:Prohibitedobjectdetection

5. Dynamic Violation Threshold – A predefined violation limit determines when an exam should be automatically terminated. If the threshold is exceeded, the system ends thetesttomaintainfairness.

6. Secure User Authentication – The platform uses JWT (JSON Web Tokens) to manage user authentication and session security, ensuring that only authorized students canaccesstheexam.

7. Code Execution Module – The system allows real-time execution of Python, JavaScript, and Java code, providing instant feedback to students while preventing unauthorizeddebuggingtools.

8. Multer for File Handling – Multer is used for efficient file management, securely handling the storage of examrelatedscreenshotsandlogsforlaterreview.

9. Express.js Backend – A Node.js-based backend powered by Express.js ensures smooth handling of exam routes, question management, and logging of cheating incidents.

10. MongoDB Database – A NoSQL MongoDB database stores essential exam details, questions, user information, and violation logs, ensuring structured and scalable data management.

Fig.7.4:ActivitylogginginDatabase

11. Redux for State Management – The system uses Reduxtomaintainacentralizedstate,efficientlymanaging user sessions, exam progress, and violation tracking acrosstheplatform.

Fig.7.5:Userauthentication.

12. Material UI for UI Components – A modern and responsiveUIisbuiltusingMaterialUI,ensuringavisually appealingandconsistentexaminterfaceforusers.

13. Snackbar Notifications – Real-time alerts are displayed using Snackbar notifications, immediately informing users and invigilators of detected violations or warnings.

14. Automatic Exam Termination – If violations surpass the allowed threshold, the system automatically ends the

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

exam, preventing further dishonest behavior and maintainingacademicintegrity.

15. Dashboard for Teachers – Teachers have access to a dedicated dashboard displaying real-time cheating logs, flagged screenshots, and violation summaries for fair assessment.

Network stability: The systemshouldassurethatit isstill functionalinvariousinternetconditions.

i) Data encryption: Allowing safe data texting and savingofuserinformation, conformingtoprivacy rules.

ii) Interface responsiveness: Making sure the user interface is self-explanatory and reliable thus allowinginvigilatorsandstudentstointeractwith theprogramsmoothly.

In addition to automated checks, real-time user testing is conducted to gather feedback from both exam proctors and students. This, in turn, suggests that besides the technical operation of the system, the user interface is exciting and easy to use. The major elements that have beentestedinclude:

Fig.7.6:TeacherDashboard

16. RESTful API Endpoints – The backend provides RESTful APIs for managing exam data, cheating logs, and user sessions, ensuring smooth and efficient data exchange.

17. Object Detection Feedback Loop – The system continuously scans for unauthorized objects every three seconds, maintaining proactive surveillance throughout theexam.

18. Secure HTTPS Connection – Data transmission between the client and server is secured using HTTPS, protecting sensitive exam information from unauthorized access.

19. Cross-Browser Compatibility – The platform is designed to function seamlessly across various web browsers, ensuring accessibility without performance issues.

20. Scalable Architecture – Built to support a large number of concurrent users, the system ensures stable andefficientperformanceevenunderhighexamloads.

8 EVALUATION AND TESTING

Regarding an AI-driven proctoring system, its assessment procedure is two-fold, referring to the measurement of both the technical capability of the system and its realworldapplicability.Thesystemischeckedbyamixtureof automated and manual tests to validate whether it operates more than one way (small tests, large exams, etc.) at best. Automated testing is used to examine functionalpartsofitasfollows:

i) Detection Accuracy: How good the AI is at catching things like eye movement anomalies or noises outside the field which will result in misconduct. Moreover, this will make sure all the violationsarecaught.

ii) Scalability and Performance: The ability of the system to support many users at a time without affectingtheperformanceofthesystem.

iii) User Privacy: Guaranteeing that the system complies with data privacy laws without compromisingtheproctoringsystem.

iv) SystemUsability:Checkingalongsidethementors how easy the technology is to use, if any changes could be applied, and how well it functions in general.Thesewillinturnhelpthesystemevolve accordingtoreal-worldconditions.

Allthisfinallycumulatesinfeedback-drivenimprovement to ensure that the system deployed at scale can be confident.Everyfindingofthetestingphasefeedsdirectly into updates to the system, including weaknesses and improvingreliabilityfortheoverallplatform.It'sarobust, scalable, and secure solution for online exam monitoring, whichensuresfairness,transparency,andusertrust.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

Fig.8.1:ActivityDiagram

Fig.8.2:SequenceDiagram

9. RESULTS, PERFORMANCE AND ANALYSIS

1. AccuracyandEfficiency

a. Averageobjectdetectiontime=30ms

b. Screenshotsavesuccessrate=98%

c. Exam termination logic success rate =100%

2. ScreenshotsCapturedandLogged

a. Number of cheating events logged per session

b. AverageFacedetectionaccuracy=95%

3. LoadandScalabilityTesting

a. Handled 100 concurrent exams with no downtime

4. DataProtection

a. JWT-basedsecureauthentication

b. HTTPSforsecuredatatransmission.

5. PrivacyHandling

a. Screenshotsareencryptedatrest

b. Logsaredeletedaftersessionexpiry.

6. ComparativeAnalysis

Feature ExistingSystems Proposedsystems

Human dependance High Minimal

Real

Automated Screenshot capture

Violation Logging Basic Detailed and Automatic Code Execution No Yes

Fig.9.1:Comparativeplot

10. CONCLUSION

An AI driven proctoring system was developed in the project to cater some of the major challenges of online exam securities. A solution that uses machine learning algorithm and does real-time video based surveillance present a better secure approach for student monitoring during exam without human being involved. The project implements multilevel detections by surveying both environmentandbehavioralconditionswhichenhancethe sedulity of proctor operation for any online examination. Since, it also considers privacy issue norms and user friendliness feature that makes it composite unique solution according to current new era trend of online education.

Moving ahead with advanced facilities like real-time feedback,multi-languagesupportandmore,inthisproject roadmap along with the collaboration will help us in achieving our final full deployment of proctoring model whichwillbethemostintuitiveone,tofacilitatethateasyto-use environment and decrease false positive rate modifications are added. Moreover, future work is also focusing on scalability and adaptability of different educational platforms and students. This project provides an extensive modern solution for online examination to

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN: 2395-0072

become a secure, time consuming environment friendly user’se-proctor.

11 REFERENCES

[1]Covid-19pandemicandonlinelearning:thechallenges andopportunities

https://www.tandfonline.com/doi/abs/10.1080/104 94820.2020.1813180

[2]ASystematicReviewonAI-basedProctoringSystems: Past,PresentandFuture

https://link.springer.com/article/10.1007/s10639021-10597-x

[3]OnlineStudentAuthentication

https://ieeexplore.ieee.org/abstract/document/9427 463

[4] A Systematic Review of Deep Learning Based Online Exam Proctoring Systems for Abnormal Student Behaviour Detection https://www.researchgate.net/profile/Saad-Abid4/publication/363432194_A_Systematic_Review_of_D eep_Learning_Based_Online_Exam_Proctoring_System s_for_Abnormal_Student_Behaviour_Detection/links/6 31c31df0a70852150e1d964/A-Systematic-Reviewof-Deep-Learning-Based-Online-Exam-ProctoringSystems-for-Abnormal-Student-BehaviourDetection.pdf

[5] The Role Of Machine Learning Algorithms In Enhancing TheAccuracyAndTheReliability OfFacial Identity, Verification In Online Assessment https://www.researchgate.net/profile/TemitopeAdetunji2/publication/383384321_The_Role_of_Machine_Lear ning_Algorithms_in_Enhancing_the_Accuracy_and_The _Reliability_of_Facial_Identity_Verification_in_Online_ Assessment/links/66ca11fb75613475fe7879e1/TheRole-of-Machine-Learning-Algorithms-in-Enhancingthe-Accuracy-and-The-Reliability-of-Facial-Identity-V

[6] Subramani, R., & Arulpandy, P. (2023). AI-Driven Proctoring System. In Engineering, Science, and Sustainability(pp.200-205).CRCPress.

[7] Awaghade, A. T., Bombe, D. A., Deshmukh, T. R., & Takawane, K. D. (2017). Online Exam Proctoring System. International Journal of Advance Engineering and Research Development (IJAERD), e-ISSN: 23484470,print-ISSN:2348-6406.

[8] Towards Data Science. (n.d.). Automating Online Proctoring Using AI. Retrieved from https://towardsdatascience.com/automating-onlineproctoring-using-ai-e429086743c8

[9] How Common is Cheating in Online Exams and did it Increase During the COVID-19 Pandemic? https://link.springer.com/article/10.1007/s10805023-09485-5