EGOFALLS
收藏资源简介:
EGOFALLS数据集是由荷兰格罗宁根大学创建的,旨在通过第一人称视角摄像机捕捉的视频来检测跌倒事件。该数据集包含10,948个视频样本,涉及14名参与者,包括老年人和年轻健康个体。数据集不仅包含RGB视频,还包括红外视频和音频记录,以适应不同光照条件下的跌倒检测。数据收集过程中,摄像机被放置在参与者的腰部和颈部等不同位置,以获取全面的视觉、运动和音频信息。该数据集的应用领域主要集中在老年人跌倒检测和相关活动识别,旨在通过多模态数据融合提高跌倒检测的准确性和可靠性。
The EGOFALLS dataset was developed by the University of Groningen in the Netherlands, with the goal of detecting fall events via videos captured by first-person perspective cameras. This dataset comprises 10,948 video samples involving 14 participants, including both elderly individuals and young healthy adults. It includes not only RGB videos but also infrared videos and audio recordings, to facilitate fall detection under varying lighting conditions. During the data collection process, cameras were mounted at different positions on participants such as the waist and neck, to obtain comprehensive visual, motion and audio information. The primary application areas of this dataset focus on elderly fall detection and related activity recognition, aiming to improve the accuracy and reliability of fall detection through multimodal data fusion.

- 1EGOFALLS: A visual-audio dataset and benchmark for fall detection using egocentric cameras格罗宁根大学 · 2023年



