GazeBase

GazeBase是一个大规模、多刺激、纵向的眼动数据集，包含来自322名大学年龄段受试者的12,334次单眼眼动记录。受试者在每次记录的连续两轮中完成七项任务，包括固定任务、水平扫视任务、随机斜向扫视任务、阅读任务、电影视频自由观看任务和视线驱动游戏任务。数据集通过EyeLink 1000眼动追踪器以1,000 Hz的采样率收集，每项任务前都进行了校准和验证以确保数据质量。由于其大量的受试者和纵向特性，GazeBase非常适合用于眼动生物识别和其他应用机器学习技术于眼动信号分析的研究假设探索。

GazeBase is a large-scale, multi-stimulus, longitudinal eye-tracking dataset comprising 12,334 monocular eye-tracking recordings from 322 university-aged participants. For each recording, participants completed seven tasks across two consecutive sessions, including fixation tasks, horizontal saccade tasks, random oblique saccade tasks, reading tasks, free viewing of movie and video tasks, and gaze-driven game tasks. The dataset was collected at a sampling rate of 1,000 Hz using an EyeLink 1000 eye-tracker, and calibration and validation were performed prior to each task to ensure data quality. Owing to its large participant cohort and longitudinal design, GazeBase is well-suited for exploring research hypotheses in eye movement biometrics and other applications of machine learning techniques to eye movement signal analysis.

搜集汇总

背景与挑战

背景概述

GazeBase是一个大规模、多刺激、纵向的眼动数据集，包含322名大学年龄段受试者的12,334次单眼眼动记录，受试者完成七项任务（如固定、扫视、阅读等），数据以1,000 Hz采样率收集并经过校准验证。由于其丰富的任务设计和纵向特性，该数据集非常适合眼动生物识别和机器学习在眼动信号分析中的应用研究。