A Comprehensive Survey of AI-Driven Smart Campus Systems for Adaptive Learning and Assessment

Abstract

Smart Campus project is a learning and management application that is meant to enhance learning experiences and operational efficiency in institutions. The system supports students, faculty, and administrators by means of academic AI, Retrieval-Augmented Generation (RAG), and data analytics. It provides adaptive learning tracks, dynamic course recommendations, and skills-based assessments, as well as productivity and engagement tools. The platform also has real-time analytics dashboard to monitor performance in academics, attendance, and resource utilization to aid the stakeholders in making effective decisions. Smart Campus responds to the deficiency of the traditional education system as it lacks individualized assistance, since the implementation of AI-based modules contributes to the academic guidance, administration, and monitoring of achievements. The outcomes of the experiments demonstrate increased efficiency and engagement, demonstrating the strengths of the platform in providing alterable data-driven sustainable campus conditions. Frequent changes according to the feedback of the users make the system relevant and approachable to the learners that have different needs and technological awareness. In education, we consider the use of technology, automated evaluation of answers and auto grading systems.

Introduction

This project proposes a Smart Campus platform that integrates Generative AI, Retrieval-Augmented Generation (RAG), multimodal models, and automated grading systems (AGS) to modernize educational assessment and management. The core objective is to create an adaptive, explainable, and scalable system that improves question generation, grading, feedback, and performance analytics.

Key Concepts and Motivation

• RAG-based LLMs enhance answer accuracy and reduce hallucinations by grounding responses in verified information.

• Large, flexible databases like MongoDB allow efficient storage and retrieval of structured and unstructured academic data, including student submissions and assessment records.

• Advances in image recognition and multimodal AI enable grading of handwritten or image-based submissions without traditional OCR.

• Automated Grading Systems (AGS) use semantic evaluation to match student responses with reference answers, providing quick, consistent, and explainable scoring.

Although AGS shows promise, concerns remain regarding bias, transparency, domain alignment, and explainability. This project aims to address these issues by integrating explainable AI principles into the platform design.

Related Work Overview

The literature review highlights:

• AI-powered collaborative learning platforms that combine automation with human oversight.

• Intelligent Tutoring Systems (ITS) and Adaptive Hypermedia Systems (AHS) that personalize learning.

• Research on Explainable AI (XAI) in education, emphasizing the need for transparent and human-centered explanations.

• Adaptive learning platforms like EduFlex, which demonstrate measurable improvements in student engagement and performance.

• RAG-based QA systems and LLM-powered assistants, where RAG often outperforms fine-tuned-only approaches.

• The evolution of AI in education from rule-based systems to modern NLP-driven intelligent systems.

These studies support the need for an integrated, explainable, AI-driven educational ecosystem.

Proposed Smart Campus System

The Smart Campus platform is a web/mobile-based academic management system powered by Generative AI and multimodal vision-language models.

Core Features:

1. Dynamic Question Generation

   • AI generates adaptive questions based on topic, difficulty, and student context.

2. Automated Semantic Grading

   • LLMs evaluate answers using semantic similarity.

   • Provides score justification and continuous feedback.

3. Multimodal Processing

   • Uses LLAMA (via Ollama) and LLaVA for text and image-based submissions.

   • Can process handwritten responses without traditional OCR.

4. Scalable Data Management

   • MongoDB stores large-scale structured and unstructured educational data.

   • Enables advanced analytics and performance monitoring.

5. Explainability and Equity

   • Clear explanation of question construction, grading logic, and score computation.

   • Addresses black-box AI concerns.

System Architecture

The platform includes three primary modules connected to an AI/ML service layer:

• Teacher Module

  • Create and assign tasks.

  • Receive AI-generated performance reports.

  • Access detailed student analytics.

• Student Module

  • Submit text or image-based answers.

  • Receive personalized feedback.

• Parent Module

  • View dashboards and real-time progress reports.

The AI layer (using frameworks such as LangChain and Hugging Face Transformers) manages question generation, answer verification, personalization, and analytics, while ensuring secure role-based access and data integrity.

Conclusion

The proposal to introduce Smart Campus is a sensible proposal to introduce AI in our classrooms in a somewhat scalable manner, uniting state-of-the-art LLM, Retrieval Augmented Generation, and multimodal vision models such as LLaVA through Ollama. In fact it puts the automation and real teaching in closer contact, allowing us to auto-create tasks, auto-assess and obtain real-time analytics on each workflow. Teachers, students, and parents find using it extremely easy due to the design allowing them to co-operate through the shared Python-FastAPI and React.js implementation. The secure database and multimodal understanding system substitutes simple OCR in such a way that the system can accurately read handwritten notes and other visual submissions and make the system scalable and adjustable to most academic environments. The findings of the experiments indicate that using Generative AI with LLaVA reduces feedback recognition errors, reduces hand-marking time, and enhances the motivation of students. Further, the extensible, modular architecture allows us to add the future AI tools predictive analytics, adaptive learning, and explainable evaluation. Ultimately, Smart Campus prepares the foundation of next-generation, AI-powered ecosystems that provide educational efficiency, transparency and personalization on both the learning and assessment side.

References

[1] M. A. Islam, M. B. Islam, M. R. Ali, M. A. Islam, G. Singh, and M. S. Hossain, “An Evaluation of AI-Enhanced Collaborative Learning Platforms, ”in Proceedings of the 2024 International Conference on Communication, Computer Sciences and Engineering (IC3SE), IEEE, 2024, pp. 207–211,doi: 10.1109/IC3SE62002.2024.10593320. [2] A. Rachha and M. Seyam,“Explainable AI in Education: Current Trends, Challenges, and Opportunities,”in Proceedings of the IEEE SoutheastCon 2023, IEEE, 2023, pp. 232–239, doi: 10.1109/SoutheastCon51012.2023.1015140. [3] R. D. Meena, S. A. Reddy, A. Bahauddin, S. T. S. Vivek, V. K. Swamy, and V. S. N. Aarya, “EduFlex: An Adaptive E-Learning Platform Using RAG and Integrated APIs,”in Proceedings of the 3rd IEEE International Conference on Knowledge Engineering and Communication Systems (ICKECS 2025), IEEE, Apr. 2025,doi: 10.1109/ICKECS65700.2025.11035512. [4] M. Ahmed, M. Darrah, A. Ashraf, Y. Adel, A. Elatrozy, B. E. Mohamed, and W. Gomaa, “CodeQA: Advanced Programming Question-Answering Using LLM Agent and Retrieval-Augmented Generation,”in Proceedings of the 6th Novel Intelligent and Leading Emerging Sciences Conference (NILES 2024), IEEE, 2024, pp. 494–499, doi: 10.1109/NILES63360.2024.10753267. [5] A. K. Dey, V. K. Chauhan, P. K. Singh, and A. Sarkar,“Exploring the Integration of Generative AI in Modern Education Systems: A Comprehensive Analysis,” in Proceedings of the IEEE International Conference on Artificial Intelligence and Quantum Computation-Based Sensor Applications (ICAIQSA 2024), IEEE, 2024, doi: 10.1109/ICAIQSA64000.2024.10882259.

Copyright

Copyright © 2026 Sourabh Mujumale, Prof. Sweta Kale, Saiashrita Panganti, Gautami Parkad, Shubham Palve. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.