Client Profile: A Japanese educational institution aiming to improve classroom dynamics by analyzing student-teacher interactions.
Challenge: Monitoring and analyzing behavior in real-time across multiple individuals in a classroom to enhance teaching effectiveness and student engagement.
YITEC’s Solution:
- Developed a system to track multiple individuals using camera feeds.
- Extracted pose estimation data, detected gaze direction, and identified specific actions for each individual.
- Implemented a scalable architecture for real-time data processing and analysis.
Core Technology:
- YOLOv8: Real-time deep learning model for human detection.
- OCSORT: Advanced algorithm for human tracking.
- RTMPose: Real-time pose estimation model.
- Kafka: Streaming platform for real-time data processing.
Key Results Delivered:
- Enabled detailed insights into classroom behavior patterns.
- Improved teacher feedback through actionable data on engagement and participation.
- Delivered a scalable solution capable of handling complex, multi-individual tracking scenarios.
Value to the Client:
- Enhanced teaching strategies based on data-driven insights.
- Improved student engagement and learning outcomes through tailored feedback.
- Future-proof system for integration with advanced educational tools.

Computer Vision Machine Learning

Classroom video-based student-teacher behavior analysis