穿戴式感測系統用於學生上課專注度之分析

Abstract

人們透過多元的方式獲取知識與經驗，但大部分的人在成長過程中，是經由就學獲取新知。而學習的過程中，專注度對於學習成效有一定程度的影響，因此，學生能否專心於課堂學習之上，成為衡量其學習成效的依據。而在過去文獻中，幾乎沒有將穿戴式裝置與動作辨識技術結合應用於課堂學習研究的。有鑑於此，本研究結合機器學習，創新地提出了頭部動作偵測、手部動作偵測及視覺焦聚辨識等模組。通過整合數種在穿戴式裝置上的感測元件 (如攝影機、加速度計和陀螺儀等) 及無線通訊功能所得到的資料，結合分類器去辨識學生在課堂上的行為。進而利用這些行為去推斷學生在不同時間區段的專注程度，由此產生學生在課堂上的專注度報表，使授課的老師或聽課的學生能瞭解上課學習時的專心情況，藉此雙方皆能做適度的調整。實驗結果顯示本方法為學生上課專注度感測提供了一個良好的解決方案，有效地對學生上課的專注度進行了分類評判。

Perception of students’ attention in a class is an important feedback to teachers. Traditionally, such information is collected manually. Wearable devices, which have received a lot of attention recently, are rarely discussed in this field. In view of this, we propose a system which integrates a head-motion module, a pen-motion module, and a visual-focus module to analyze students’ attention levels in class. These modules capture external information via cameras, accelerometers, and gyroscopes placed on wearable devices to recognize students’ behaviors. From these recognized behaviors, we then infer their attention levels under varied time periods via a rule-based approach and a data-driven approach. The former infers the student’s attention state by user-defined rules, while the latter infers that by the implied information hidden in the data. The outcome of this research may greatly help teaching and learning efficiency in class. As far as we know, this is the first work attempting to address attention inference in class in this direction. Extensive experimental results have been collected, which show great potential in understanding real-time attention.