Panonut360 全景视频头部和眼睛追踪数据集

Panonut360 是由华中科技大学的研究团队采集并推出的一个以全景视频为对象的用户头部和眼睛追踪数据集。此数据集涵盖了 50 名用户（男女各占一半）在运用 HTC VIVE Pro HMD 观看 15 种不同类型（像自然风光、动画片、体育赛事、故事片等）的 4K 高清全景视频时所产生的头部和眼睛动作，视频的观看时长超过 60 个小时，细致地记录了参与者的视口以及注视点的位置信息。 Panonut360 能够应用于视点预测、显著性检测等任务，助力研究人员开发全景视频的自适应流系统，促进全景视频技术的进步。

Panonut360 is a head and eye tracking dataset focused on panoramic videos, collected and released by the research team from Huazhong University of Science and Technology. This dataset includes head and eye movement data from 50 participants with an equal split of males and females, collected while they watched 15 different categories of 4K ultra-high-definition panoramic videos (such as natural scenery, animations, sports events, feature films, etc.) using HTC VIVE Pro HMDs. The total cumulative viewing duration exceeds 60 hours, and it meticulously records the viewport and gaze point positions of the participants. Panonut360 can be applied to tasks including viewpoint prediction and saliency detection, helping researchers develop adaptive streaming systems for panoramic videos and advancing the development of panoramic video technologies.

搜集汇总

背景与挑战

背景概述

Panonut360是一个由华中科技大学研究团队推出的全景视频头部和眼睛追踪数据集，包含50名用户观看15种不同类型全景视频时的追踪数据，视频观看时长超过60小时，数据量为1.3GB。该数据集适用于视点预测和显著性检测等任务，支持全景视频自适应流系统的开发。