OpenEDS2020

OpenEDS2020 是在受控照明下以 100 Hz 的帧速率捕获的眼睛图像序列数据集，使用安装有两个同步的面向眼睛的摄像头的虚拟现实头戴式显示器。该数据集匿名删除参与者的任何个人身份信息，由 80 名不同外表的参与者组成，执行多项凝视引发任务，并分为两个子集：1) 凝视预测数据集，多达 66,560 个序列，包含 550,400 只眼睛- 图像和相应的凝视矢量，旨在促进时空凝视估计和预测方法的研究； 2) 眼部分割数据集，由 200 个以 5 Hz 采样的序列组成，包含多达 29,500 张图像，其中 5% 包含语义分割标签，旨在鼓励使用时间信息将标签传播到连续帧。

OpenEDS2020 is a dataset of eye image sequences captured at 100 Hz frame rate under controlled illumination, utilizing a virtual reality head-mounted display fitted with two synchronized eye-facing cameras. This dataset anonymizes all personally identifiable information (PII) of its participants, and comprises data from 80 participants with diverse appearances who completed multiple gaze-eliciting tasks. It is divided into two subsets: 1. Gaze Prediction Dataset: Containing up to 66,560 sequences with a total of 550,400 eye images and their corresponding gaze vectors, this subset is designed to advance research on spatiotemporal gaze estimation and prediction methods. 2. Eye Segmentation Dataset: Consisting of 200 sequences sampled at 5 Hz with up to 29,500 total images, 5% of which are annotated with semantic segmentation labels. This subset aims to promote the propagation of labels across consecutive frames using temporal information.

搜集汇总

背景与挑战

背景概述

OpenEDS2020是一个由80名参与者匿名组成的眼睛图像序列数据集，在受控照明下以100 Hz帧率通过虚拟现实头戴式显示器捕获。它包含两个子集：凝视预测数据集用于时空凝视估计和预测研究，眼部分割数据集则用于基于时间信息的语义分割标签传播。