Chapter 4
Structural Design
4.1 Use Case Diagram
4.2 Data Flow Diagram
Chapter 5
Limitations and Future Work
5.1 Limitations:
Handwriting recognition accuracy: Spelling mistakes, inconsistent handwriting, and other factors
can all reduce the handwriting recognition system's accuracy. As a result, it's possible that the AI
system won't be able to appropriately identify and mark responses.
Limited scope: The criteria and rules that have been encoded into the AI system are the only
ones that can be used to mark replies. It might not be able to take into account other elements
that a human marker might, including the answer's applicability to the question or general
coherence.
Ethics and fairness: If students are not given the chance to contest the findings, there may be
questions about how fair it is to use an AI system to grade examinations. Moreover, there can be
issues with the application of personal data and privacy.
Hardware requirements: It takes a lot of computational power and resources, including specialist
hardware like GPUs (Graphics Processor Units) and other high-performance computing devices,
to run an AI-based attendance system using CNN.
Privacy concerns: Due to the gathering and storage of biometric data, using face recognition
technology for attendance presents privacy issues. This can be particularly delicate if the
technology is utilized in a situation where people would not want their biometric data to be
gathered and retained, such as a workplace or school.
Environmental conditions: An AI-based attendance system using CNN may function differently
depending on the lighting, shadows, and reflections of the surrounding area. If the person is
wearing accessories like glasses or hats, or if the illumination is poor, the system might not be
able to reliably identify faces.
Maintenance and updates: To keep it accurate and current, an AI-based attendance system with
CNN needs to get frequent upkeep and upgrades. This can be expensive and time-consuming,
and it could be necessary to hire experts.
Cultural differences: An AI-based attendance system using CNN may encounter difficulties due
to cultural variations, as various cultures have very varied norms for appropriate facial
expressions and hand gestures. This may lead to incorrect attendance records and misreading of
face characteristics.
5.2 Future Work
1. Multilingual support: Future research can concentrate on creating models that can read
handwriting in a variety of languages, expanding the technology's usability and setting-
specific application.
2. Natural language processing: Future research might investigate how to use natural
language processing to give a more accurate and nuanced examination of written
responses.
3. Learning analytics: Future research can look into how learning analytics can be used to
give insights into student performance, such as detecting frequent errors or problem
regions.
4. Mobile compatibility: Future work might concentrate on creating mobile applications that
allow students or workers to record their attendance using their mobile devices,
increasing accessibility and convenience.
5. Real-time monitoring: Future research can concentrate on creating methods for tracking
attendance in real-time that can spot abnormalities like students trying to game the
system or proxy attendance.
6. Robustness: Future research can look into methods to make the system more dependable
and robust, including by implementing redundancy and error-correcting systems.
7. Adapting to different exam formats: Future research can concentrate on creating
algorithms that can read handwriting in various exam forms, such multiple-choice tests or
short response questions.
8. Scaling to large datasets: Future research can look into ways to scale the system to handle
enormous datasets, for creating cloud-based or distributed computing systems.
9. Privacy-preserving techniques: Future research can look into ways to secure sensitive
information about people's privacy, including employing differential privacy or other
privacy-enhancing technologies.
10. Adapting to different environments: Future research can concentrate on creating models
that can identify faces in various lighting scenarios or with various expressions, which
will enhance the accuracy and dependability of the system.
Chapter 6
Conclusion
In conclusion, the incorporation of AI into the classroom has the potential to revolutionize the
current system of instruction by offering more individualized and interactive learning
opportunities. AI-based classroom systems can assist instructors in better understanding the
unique needs of each student, providing specialized learning materials, and monitoring the
development of each student. In addition, AI-based classroom solutions can aid in automating
administrative activities like attendance tracking and grading, freeing up instructors to devote
more time to instruction and student feedback. This may make the educational system more
effective and lighten the pressure on instructors. However, there are significant difficulties and
restrictions associated with the adoption of AI-based classroom systems, including the
requirement for accurate system calibration, privacy issues, and the possibility for bias in AI
algorithms. Thus, it is essential to create AI-based teaching systems that are open, objective, and
considerate of students' privacy. The learning experience for both students and teachers might be
improved by using AI in the classroom. Teachers may gain insights into student performance to
help them deliver more effective training, and students can get individualized feedback and
learning materials that are tailored to their particular requirements thanks to AI-powered
solutions. The traditional educational system has difficulties including teacher workload and
administrative responsibilities, which can be addressed with the employment of AI in
classrooms. Teachers may concentrate more on instructing and giving feedback to students by
automating duties like grading and attendance control.
In conclusion, introducing AI into the classroom has the potential to completely transform the
way that education is delivered, but it must be done so wisely and with an eye toward making
sure that the advantages exceed the drawbacks.
References
1. Samanta, S., & Das, N. (2019). Handwriting Recognition and Marking System Using
Deep Learning Techniques. International Journal of Computer Science and Mobile
Computing, 8(6), 46-52.
2. Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). "Why should i trust you?": Explaining
the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international
conference on knowledge discovery and data mining (pp. 1135-1144).
3. Nandakumar, R., Velumurugan, S., & Jaganathan, S. (2019). Automated Handwriting
Recognition and Evaluation System Using Deep Learning. In Proceedings of the 2019
IEEE International Conference on Communication and Signal Processing (ICCSP) (pp.
1113-1117).
4. Haque, M. N., Roy, N., & Hossain, M. A. (2021). An efficient automated attendance
system using deep learning for online education. Journal of Ambient Intelligence and
Humanized Computing, 12(8), 8417-8431.
5. Chiang, C. Y., & Huang, Y. M. (2019). Deep Learning Based Automatic Attendance
System. In Proceedings of the 2019 IEEE 6th International Conference on Industrial
Engineering and Applications (ICIEA) (pp. 120-124).
6. Abdur Rahman, M., Islam, M. T., & Hossain, M. (2021). An Automated Attendance
System Using Face Recognition for Academic Institutions. In 2021 7th International
Conference on Computing and Artificial Intelligence (ICCAI) (pp. 83-88).
7. Bhattacharya, U., Kumar, R., & Mandal, D. (2020). Automatic Answer Evaluation in
Engineering Exams using Deep Learning. In 2020 11th International Conference on
Computing, Communication and Networking Technologies (ICCCNT) (pp. 1-7).
8. Saha, S., & Das, S. (2020). Automatic Handwritten Short Answer Grading System using
Convolutional Neural Network. In 2020 10th International Conference on Cloud
Computing, Data Science & Engineering - Confluence (pp. 347-353).
9. Adesope, O., & Zhou, Y. (2020). Machine learning applications in education: A review
of recent advancements and future prospects. Journal of Educational Technology &
Society, 23(3), 1-14.
10. Raj, M., Khare, N., & Singh, M. (2020). Automated Attendance System using Deep
Learning for Online Learning Platform. In Proceedings of the 3rd International
Conference on Communication, Computing and Electronics Systems (ICCCES 2020)
(pp. 798-803).