Student Progress Monitoring System Using Deep Learning

Abhishek K

PG Department of Computer Applications JSS College of Arts, Commerce and Science Mysuru, India

Mr. Ravi Kumar V G

Head of the Department,

PG Department of Computer Applications JSS College of Arts, Commerce and Science Mysore, India

Abstract — Monitoring student academic performance is a critical function of educational institutions, yet traditional approaches remain largely manual, fragmented, and reactive. This paper presents a Student Progress Monitoring System that integrates Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) networks to automate the analysis and prediction of student academic outcomes. The system ingests academic parameters such as previous semester percentage, internal assessment marks, attendance percentage, and academic activity scores, and classifies students into performance categories of Excellent, Average, and At Risk. Optical Character Recognition (OCR), implemented using OpenCV and Tesseract OCR, is incorporated to automatically extract marks from uploaded mark-sheet images, substantially reducing manual data-entry effort. A voice-enabled academic chatbot further supports quick, conversational access to student records, analytics, and prediction outcomes. The system was implemented using Flask, TensorFlow, and Keras with a SQLite/MySQL backend, and offers department-wise analytics, semester-wise reporting, subject-wise performance tracking, and role-based access for administrators and faculty. Experimental evaluation shows that the hybrid ANN-LSTM model achieves an accuracy of 96% in predicting student performance, demonstrating its effectiveness in supporting early identification of at-risk students and enabling data-driven academic intervention.

Keywords — Student Performance Prediction; Deep Learning; Artificial Neural Network; Long Short-Term Memory; Optical Character Recognition; Educational Data Mining; Academic Analytics; Chatbot.