SEENIC
收藏资源简介:
SEENIC数据集是首个专为航天器姿态估计设计的事件相机数据集,由法国蔚蓝海岸大学LEAT实验室创建。该数据集包含5415个事件帧,分为训练集和测试集,数据来源于哈勃太空望远镜的模拟和真实事件帧。数据集通过V2E软件将RGB图像转换为事件流,但由于模拟过程中的局限性,事件流的时间连续性和极性信息存在不足。SEENIC数据集主要用于评估事件驱动的神经网络在航天器姿态估计中的性能,旨在解决航天器在轨服务(OOS)和主动碎片清除(ADR)任务中的姿态估计问题,特别是在资源受限的嵌入式系统中实现高效、低功耗的姿态估计。
The SEENIC dataset is the first event camera dataset tailored specifically for spacecraft attitude estimation, developed by the LEAT Laboratory at Université Côte d'Azur, France. This dataset contains 5415 event frames, divided into training and test sets, with data sourced from both simulated and real event frames of the Hubble Space Telescope. The dataset converts RGB images into event streams via the V2E software; however, due to limitations in the simulation process, the event streams have deficiencies in temporal continuity and polarity information. The SEENIC dataset is primarily used to evaluate the performance of event-driven neural networks in spacecraft attitude estimation, aiming to address the attitude estimation challenges in spacecraft on-orbit servicing (OOS) and active debris removal (ADR) missions, particularly to achieve efficient, low-power attitude estimation in resource-constrained embedded systems.




