以人眼辨識偵測上課是否專心

Abstract

本系統是以Python開發一套以學生的眼睛來判別是否專心上課的偵測程式，主要目的是希望能夠有效掌控學生在課堂上有專心上課。這個程式將透過提供學生若有出現不專心的狀況時，將會即時拍攝其照片並傳送至授課教師，協助監督學生的上課狀態，同時提高學生的學習成果。 在這個程式中，我們利用了Webcam的攝影功能和數種Python影像辨識套件，包括OpenCV和dlib等，來追蹤學生眼睛的位置，並分析瞳孔與頭部的相對移動幅度和頻率，從而評估學生的專心程度。同時，本系統還加入了簡易的防偽機制，以防範常見的相片攻擊(Photo Attack)和俗稱「貼假眼睛」的化妝攻擊(Makeup Attack)，來確保程式準確又可靠。 經過實測，本系統的程式能夠有效偵測學生的不專心狀態及方法，並提供準確的分心相片紀錄給教授作為參考。本系統程式的應用範圍非常廣泛，可以靈活運用於任何具備webcam的電腦裝置，並且特別適用於大學電腦教室中，進而達到促進學生學習成果的目的。

This report aims to develop a Python program objectively assess students' concentration levels to solve student distraction problems in Taiwanese university classrooms. The program assists in monitoring students' engagement during class by providing professors with photographic records of students' distracted moments while enhancing students' learning outcomes. In this program, we leverage the webcam functionality and various Python image recognition libraries such as OpenCV and dlib to track the position of students' eyes. We analyze the relative movements and frequencies of the pupils and head to evaluate students' level of concentration. Additionally, we have implemented a simple anti-fraud mechanism to prevent common photo attacks and makeup attacks through eye stickers, ensuring the reliability and accuracy of the program. Through several testing, our program effectively detects students' states of distraction and provides accurate photographic records for professors as references. The application scope of this program is highly versatile since it can be creatively used on any computer with a webcam, particularly in computer classrooms, thereby promoting students' learning outcomes.