N-Omniglot

N-Omniglot是由中国科学院自动化研究所脑启发认知智能实验室创建的大型神经形态数据集，专注于时空稀疏小样本学习。该数据集基于动态视觉传感器，包含1623个手写文字类别，每类仅有20个样本。N-Omniglot通过记录笔画的时间顺序信息，为脉冲神经网络提供了丰富的时空特征，适用于开发和评估小样本学习算法。数据集的创建过程涉及将原始笔画数据转换为视频，再通过神经形态相机捕捉事件记录。N-Omniglot的应用领域主要集中在脉冲神经网络的小样本学习，旨在解决传统数据集在时空信息表达上的不足。

N-Omniglot is a large-scale neuromorphic dataset developed by the Brain-Inspired Cognitive Intelligence Laboratory, Institute of Automation, Chinese Academy of Sciences, which focuses on spatio-temporal sparse few-shot learning tasks. This dataset is built upon dynamic vision sensors, encompassing 1623 handwritten character categories, with only 20 samples per category. By recording the temporal sequence information of writing strokes, N-Omniglot provides rich spatio-temporal features for spiking neural networks (SNNs), and is suitable for developing and evaluating few-shot learning algorithms. The construction process of N-Omniglot involves converting raw stroke data into video footage, then capturing event records via neuromorphic cameras. The primary application scenarios of N-Omniglot center on few-shot learning for spiking neural networks, aiming to address the limitations of traditional datasets in expressing spatio-temporal information.