Synthetic Microsaccade Dataset
收藏资源简介:
该数据集由爱尔兰国立高威大学工程学院 C3I Imaging Lab 的研究人员创建,包含 175,000 个注解序列样本,模拟了真实水平的眼睛运动,共分为七个幅度范围,从 0.5° 到 2.0°。数据集使用 Blender 渲染 RGB 帧生成,并通过事件模拟器 v2e 转换为事件流,以支持神经形态视觉模型的评估。数据集旨在提供一个受控的环境,以系统评估和重现神经形态视觉模型,并支持未来的研究。
This dataset was developed by researchers at the C3I Imaging Lab, School of Engineering, National University of Ireland Galway. It comprises 175,000 annotated sequence samples that simulate realistic eye movements, grouped into seven amplitude ranges spanning from 0.5° to 2.0°. The dataset was created by rendering RGB frames via Blender, then converted into event streams using the event simulator v2e to facilitate the evaluation of neuromorphic vision models. Its purpose is to provide a controlled environment for the systematic evaluation and reproduction of neuromorphic vision models, as well as to support future research endeavors.




