OpenEDS

OpenEDS，全称Open Eye Dataset，是安装在VR头盔中的眼控摄像头收集的眼球图像的大规模数据集。Facebook希望通过这个数据集的开放，推动虚拟现实领域的技术发展。

数据集由从具有像素级注释 (巩膜，虹膜，瞳孔) 的152对象的眼睛视频中随机选择的12,759图像组成。该数据集还包含251,690未标记的眼睛图像。眼图像采用红外相机采集，与实验中的数据采集格式有较高的匹配度

OpenEDS, whose full name is Open Eye Dataset, is a large-scale eye image dataset collected by eye-tracking cameras mounted on VR headsets. Facebook aims to promote technological progress in the field of virtual reality by open-sourcing this dataset. The dataset consists of 12,759 randomly selected images from eye videos of 152 subjects, with pixel-level annotations for sclera, iris and pupil. It also contains 251,690 unlabeled eye images. These eye images are captured using infrared cameras, which have high compatibility with the data acquisition formats employed in related experiments.

搜集汇总

背景与挑战

背景概述

OpenEDS是一个由Facebook于2021年发布的大规模眼球图像数据集，包含12,759张带有像素级注释（巩膜、虹膜、瞳孔）的图像和251,690张未标记图像，所有图像均通过VR头盔中的红外眼控摄像头采集。该数据集旨在促进虚拟现实领域的技术发展，特别是眼动追踪和图像分割相关研究。