MOEMO - Online Learning Analytics Dashboard

Real-time emotion and attention detection system for online learning environments

Overview

Developed at Hosei University, MOEMO is a real-time learning analytics dashboard designed to detect and visualize student engagement, emotion, and attention in online learning environments.

Key Features

  • Engagement Detection: Built using MTCNN (Multi-task Cascaded Convolutional Networks) for face detection
  • Emotion Recognition: Implemented Mini-Xception model for real-time emotion classification
  • Attention Tracking: Monitored student attention levels during online sessions
  • Dashboard Visualization: Interactive dashboard for educators to monitor class engagement

System Architecture

Video Stream → Face Detection (MTCNN) → Emotion Recognition (Mini-Xception)
                                              ↓
                                    Dashboard Visualization

Technologies Used

  • Face Detection: MTCNN
  • Emotion Model: Mini-Xception
  • Backend: Python
  • Visualization: Web-based dashboard
  • Framework: TensorFlow/Keras

Publications

This work contributed to multiple publications:

  1. “Students’ Emotion extraction and visualization for engagement detection in online learning” - KES 2021
  2. “A Real-Time Learning Analytics Dashboard for Automatic Detection of Online Learners’ Affective States” - Sensors 2023

Impact

The system helps educators understand student engagement in remote learning settings, enabling timely interventions to improve learning outcomes.