ES-ImageNet

ES-ImageNet是由清华大学脑启发计算研究中心开发的大型事件流分类数据集，旨在通过软件生成的方式，将传统的计算机视觉数据集ILSVRC2012转换为适用于脉冲神经网络（SNNs）的事件流格式。该数据集包含约130万个样本，分布在1000个类别中，是目前最大的神经形态分类数据集。通过ODG算法，数据集实现了从静态图像到事件流的转换，减少了数据冗余，提高了生成速度。ES-ImageNet适用于研究SNNs在处理稀疏和时间数据时的能力，为神经形态视觉研究提供了一个新的、大规模的基准数据集。

ES-ImageNet is a large-scale event stream classification dataset developed by the Brain-Inspired Computing Research Center of Tsinghua University. It aims to convert the traditional computer vision dataset ILSVRC2012 into event stream format suitable for spiking neural networks (SNNs) via software generation. This dataset contains approximately 1.3 million samples distributed across 1000 categories, making it the largest neuromorphic classification dataset to date. The conversion from static images to event streams is implemented through the ODG algorithm, which reduces data redundancy and improves generation speed. ES-ImageNet is applicable for researching the capabilities of SNNs in processing sparse and temporal data, providing a new large-scale benchmark dataset for neuromorphic vision research.