A Dual-Camera Lecture Capturing System using YOLOv8, Whisper Transcription, and Adaptive Threshold Binary Image Enhancement

This study developed an advanced lecture capturing system designed to enhance student access to classroom content by combining real-time video capture, automated transcription, and image processing. The system features a dual-camera setup: one camera employs YOLOv8 for accurate, real-time tracking of the instructor’s movements, while the second captures the lecture space, focusing on whiteboard content. To facilitate post-lecture review, Whisper was integrated to transcribe audio into a detailed, searchable text format, achieving 99% accuracy. Additionally, Adaptive Thresholding Binary was applied to improve the clarity of captured whiteboard images, ensuring text readability and enhancing visual quality by increasing brightness and sharpness. All lecture materials, including the video recording, transcribed text in PDF, and enhanced board images, were uploaded to a cloud storage platform, providing convenient and organized access for students. Tested in real-world classroom conditions, the system demonstrated exceptional accuracy, reliability, and functionality, consistently tracking the instructor within a 4-meter range, producing clear transcriptions, and delivering high-quality visuals. This lecture capture system presents a robust, accessible solution for students to review lecture content, promoting better learning outcomes and supporting those who may miss in-person classes.