EgoCampus
收藏资源简介:
EgoCampus是由罗格斯大学创建的以行人为中心的户外导航数据集,旨在研究人类在真实世界导航中的视觉注意力。该数据集包含来自82名参与者的32小时多模态视频数据(约350万帧),覆盖25条独特的校园路径,总长度达6公里。数据通过Meta的Project Aria眼镜采集,集成了高分辨率RGB视频、眼动追踪坐标、惯性测量单元(IMU)和GPS等传感器数据。数据集经过严格的隐私处理,包括人脸模糊化。该数据集为研究行人空间注意力提供了宝贵资源,并有望应用于导航系统开发,以模拟人类注意力机制。
EgoCampus is a pedestrian-centric outdoor navigation dataset developed by Rutgers University, designed to investigate human visual attention during real-world navigation. This dataset contains 32 hours of multimodal video data (approximately 3.5 million frames) from 82 participants, covering 25 unique campus routes with a total length of 6 kilometers. The data was collected using Meta's Project Aria smart glasses, and integrates high-resolution RGB video, eye-tracking coordinates, inertial measurement unit (IMU), GPS and other sensor data. The dataset has undergone rigorous privacy processing, including facial blurring. This dataset provides a valuable resource for research on pedestrian spatial attention, and holds promise for applications in navigation system development to simulate human attention mechanisms.
EgoCampus 数据集概述
数据集状态
- 数据集与代码即将发布。
备注
- 根据提供的README文件内容,目前尚无关于数据集规模、内容、格式、获取方式及使用许可等具体信息。

- 1EgoCampus: Egocentric Pedestrian Eye Gaze Model and Dataset罗格斯大学 · 2025年



