Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AI-assisted Gaze Detection for Proctoring Online Exams

Yong-Siang Shih, Zach Zhao, Chenhao Niu, Bruce Iberg, James Sharpnack, Mirza Basim Baig

TL;DR

An AI-assisted gaze detection system is presented, which allows proctors to navigate between different video frames and discover video frames where the test taker is looking in similar directions, and enables proctors to work more effectively to identify suspicious moments in videos.

Abstract

For high-stakes online exams, it is important to detect potential rule violations to ensure the security of the test. In this study, we investigate the task of detecting whether test takers are looking away from the screen, as such behavior could be an indication that the test taker is consulting external resources. For asynchronous proctoring, the exam videos are recorded and reviewed by the proctors. However, when the length of the exam is long, it could be tedious for proctors to watch entire exam videos to determine the exact moments when test takers look away. We present an AI-assisted gaze detection system, which allows proctors to navigate between different video frames and discover video frames where the test taker is looking in similar directions. The system enables proctors to work more effectively to identify suspicious moments in videos. An evaluation framework is proposed to evaluate the system against human-only and ML-only proctoring, and a user study is conducted to gather feedback from proctors, aiming to demonstrate the effectiveness of the system.

AI-assisted Gaze Detection for Proctoring Online Exams

TL;DR

An AI-assisted gaze detection system is presented, which allows proctors to navigate between different video frames and discover video frames where the test taker is looking in similar directions, and enables proctors to work more effectively to identify suspicious moments in videos.

Abstract

For high-stakes online exams, it is important to detect potential rule violations to ensure the security of the test. In this study, we investigate the task of detecting whether test takers are looking away from the screen, as such behavior could be an indication that the test taker is consulting external resources. For asynchronous proctoring, the exam videos are recorded and reviewed by the proctors. However, when the length of the exam is long, it could be tedious for proctors to watch entire exam videos to determine the exact moments when test takers look away. We present an AI-assisted gaze detection system, which allows proctors to navigate between different video frames and discover video frames where the test taker is looking in similar directions. The system enables proctors to work more effectively to identify suspicious moments in videos. An evaluation framework is proposed to evaluate the system against human-only and ML-only proctoring, and a user study is conducted to gather feedback from proctors, aiming to demonstrate the effectiveness of the system.
Paper Structure (9 sections, 1 equation, 4 figures)

This paper contains 9 sections, 1 equation, 4 figures.

Table of Contents

  1. Introduction
  2. System Overview
  3. Evaluation Framework
  4. Human-ML Complementarity
  5. Proposed Experiments
  6. User Study
  7. Conclusion
  8. Limitations
  9. Acknowledgments

Figures (4)

  • Figure 1: The user interface of the gaze system allows proctors to navigate to different video frames using the video player timeline. The points on the gaze plot represent the gaze direction of each frame. When proctors select regions on the gaze plot, the corresponding frames on the timeline would be highlighted. The gaze direction of the current frame is colored in red in the gaze plot.
  • Figure 2: Each frame of the test session is shown as a point in the gaze plot, and the position of each point represents the gaze direction in each frame. The current frame's location is colored in red.
  • Figure 3: The video player timeline allows proctors to navigate to specific moments by clicking on the desired timestamp. Timestamps where gaze predictions fall within the selected region of the gaze plot are highlighted in blue. The space on top of the timeline can be used to show other notable events. The white bar below is used to select a specific time interval to zoom into.
  • Figure 4: Survey questions: (Q1) I felt comfortable utilizing the tool, (Q2) I felt confident that the tool was providing me with correct information, (Q3) I felt the documentation/videos provided allowed me to easily understand how to use the tool, (Q4) I didn't have difficulty interpreting or understanding any visual elements of the tool, (Q5) I found it easy to incorporate the tool in my normal proctoring processes.