GazeCapture 眼动追踪数据集

无论是科研还是商用，眼动追踪技术是社会许多领域的重要工具。这一技术尽管有如此广泛的应用范围，但是还未能广泛地普及。为了助力眼动追踪技术的研究与推广，GazeCapture 数据集应用而生。作为全球首个用于眼动追踪技术的大型数据集，GazeCapture 包含了超过 1,450 位志愿者的 250 万帧数据。通过对该数据集的应用，用于眼动追踪研究的卷积神经网络 —— iTracker 得到了充分训练，与之前的方法相比，误差大大减少。

Eye-tracking technology serves as a critical tool across numerous domains in society, whether for scientific research or commercial applications. Despite its broad application scope, this technology has not yet achieved widespread adoption. To facilitate the research and popularization of eye-tracking technology, the GazeCapture dataset was developed. As the world's first large-scale dataset for eye-tracking technology, GazeCapture contains over 2.5 million frames of data from more than 1,450 volunteers. By utilizing this dataset, the convolutional neural network for eye-tracking research, iTracker, was sufficiently trained, resulting in a significant reduction in error compared to previous methods.

搜集汇总

背景与挑战

背景概述

GazeCapture是全球首个用于眼动追踪技术的大型数据集，包含超过1,450位志愿者的250万帧数据，旨在助力眼动追踪技术的研究与推广。该数据集通过训练卷积神经网络iTracker，显著减少了眼动追踪的误差，适用于图像识别、深度学习等计算机视觉领域。