Nascimento Jade Antunes, Bazán Paulo Rodrigo, Azevedo Neto Raymundo Machado de, Silva Edilene Santos, Soares Daniela Arruda, Balardin Joana Bisol, Amaro Junior Edson
Brain Istitute, Hospital Israelita Albert Einstein, São Paulo, SP, Brazil.
Multidisciplinary Health Institute, Universidade Federal da Bahia, Vitória da Conquista, BA, Brazil.
Einstein (Sao Paulo). 2025 Apr 7;23:eAO0911. doi: 10.31744/einstein_journal/2025AO0911. eCollection 2025.
Webcam-based eye tracking offers a practical approach for monitoring reading behavior in classroom environments. Nascimento et al. demonstrated the feasibility of using the Webcam-based eye tracking to observe comic reading patterns among children and showed that it effectively captures fixation time and reading dynamics. Despite the data quality challenges, Webcam-based eye tracking provided an accuracy comparable to that of infrared-based systems.
To evaluate the feasibility and accuracy of webcam-based eye tracking for monitoring comic books reading behavior in a real-world classroom setting.
We tested the feasibility of using the webcam-based eye tracking in a children's classroom experiment (n=22), observing reading patterns as students engaged with a comic strip. Qualitative quality control was implemented to exclude low-quality data. Fixation dwell time was measured to determine whether specific areas of the image received more attention. Additionally, a validation test was conducted (n=3) to assess the accuracy and precision of webcam-based eye tracking compared with an infrared-based eye-tracking system. The accuracy and precision were evaluated, with lower values indicating better performance.
During the comic reading task, webcam-based eye tracking effectively captured fixation dwell time, showing that students read the left side significantly faster (2.916 words/s, SD=0.705) compared to the right side (1.962 words/s, SD=0.450, p≤0.001). The validation test showed that webcam-based eye tracking had an average accuracy of 11.581% and a precision of 3.058%, whereas the infrared-based system recorded an accuracy of 11.290% and a precision of 1.264%. Although webcam-based eye tracking demonstrated a slightly lower precision (bias=-1.794, 95%CI=-3.055 to -0.532), no significant difference was observed in accuracy between the two systems.
■ Webcam-based eye tracking demonstrated feasibility for monitoring reading behavior in classrooms.
■ Webcam-based eye tracking captured fixation times, showing differing reading speeds across comic sections.
■ Webcam-based eye tracking showed accuracy comparable to infrared-based systems, despite lower precision.
■ Limitations resulting in data loss underscore the need for methodological improvements.
基于网络摄像头的眼动追踪为监测课堂环境中的阅读行为提供了一种实用方法。纳西门托等人证明了使用基于网络摄像头的眼动追踪来观察儿童漫画阅读模式的可行性，并表明它能有效捕捉注视时间和阅读动态。尽管存在数据质量挑战，但基于网络摄像头的眼动追踪提供的准确性与基于红外的系统相当。
评估在现实课堂环境中基于网络摄像头的眼动追踪用于监测漫画阅读行为的可行性和准确性。
我们在一项儿童课堂实验（n = 22）中测试了使用基于网络摄像头的眼动追踪的可行性，观察学生阅读连环漫画时的阅读模式。实施了定性质量控制以排除低质量数据。测量注视停留时间以确定图像的特定区域是否受到更多关注。此外，进行了一项验证测试（n = 3），以评估基于网络摄像头的眼动追踪与基于红外的眼动追踪系统相比的准确性和精确性。对准确性和精确性进行了评估，值越低表明性能越好。
在漫画阅读任务期间，基于网络摄像头的眼动追踪有效地捕捉了注视停留时间，表明学生阅读左侧的速度（2.916词/秒，标准差 = 0.705）明显快于右侧（1.962词/秒，标准差 = 0.450，p≤0.001）。验证测试表明，基于网络摄像头的眼动追踪的平均准确率为 %，精确率为3.0%，而基于红外的系统记录的准确率为11.290%，精确率为1.264%。尽管基于网络摄像头的眼动追踪的精确率略低（偏差 = -1.7%，95%置信区间 = -3.055至 -0.532），但两个系统在准确性上没有观察到显著差异。
■ 基于网络摄像头的眼动追踪证明了在课堂上监测阅读行为的可行性。
■ 基于网络摄像头的眼动追踪捕捉到了注视时间，显示出漫画各部分的阅读速度不同。
■ 基于网络摄像头的眼动追踪显示出与基于红外的系统相当的准确性，尽管精确率较低。
■ 导致数据丢失的局限性突出了方法改进的必要性。 （注：原文中部分数据百分比处有缺失，已按原文格式保留）
