mEBAL2
收藏资源简介:
mEBAL2是由马德里自治大学Biometrics and Data Pattern Analytics Laboratory创建的,是目前最大的眼眨数据库。该数据集包含来自180名不同学生的21,100个图像序列,总计超过200万张标记图像,用于训练和评估基于RGB和近红外(NIR)图像的眼眨检测。mEBAL2通过多种传感器,包括两个NIR和一个RGB摄像头,捕捉执行任务时的面部表情,以及使用脑电图(EEG)带获取用户的认知活动和眨眼事件。此外,该数据集还探索了在真实环境中使用NIR光谱改善眨眼检测的潜力,并评估了新的数据驱动方法。mEBAL2的应用领域包括在线教育中的注意力水平估计和安全性提升,旨在通过分析眨眼模式来优化在线学习体验和评估的准确性。
mEBAL2 was developed by the Biometrics and Data Pattern Analytics Laboratory at the Universidad Autónoma de Madrid, and it is currently the largest publicly available eye blink detection dataset to date. This dataset comprises 21,100 image sequences from 180 distinct students, totaling over 2 million annotated images, and is designed for training and evaluating RGB and near-infrared (NIR) image-based blink detection systems. mEBAL2 captures facial expressions during task performance via multiple sensors, including two NIR cameras and one RGB camera, and collects users' cognitive activities and blink events using electroencephalography (EEG) headsets. Furthermore, this dataset explores the potential of leveraging NIR spectra to enhance blink detection in real-world scenarios, and evaluates novel data-driven methods. Application scenarios of mEBAL2 include attention level estimation and safety enhancement in online education, which aims to optimize online learning experiences and the accuracy of assessments by analyzing blink patterns.

- 1mEBAL2 Database and Benchmark: Image-based Multispectral Eyeblink Detection马德里自治大学 · 2024年



