ILSVRC 2014
收藏arXiv2025-09-30 收录
下载链接：
http://www.image-net.org/challenges/LSVRC/2014/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了来自2014年ILSVRC数据集的训练图像和验证图像，其中训练图像是随机选取的，并指定了一定数量的图像用于测试。此外，该数据集用于评估由模型生成的对抗性干扰的欺骗率。在规模上，每个类别的训练图像有10,000张，验证图像有50,000张。该数据集的任务是对抗性干扰的生成与评估。
This dataset comprises training and validation images derived from the 2014 ILSVRC dataset. The training images were randomly selected, with a subset designated for testing purposes. Furthermore, this dataset is utilized to evaluate the deception rate of adversarial perturbations generated by machine learning models. In terms of scale, each category contains 10,000 training images and 50,000 validation images. The core task of this dataset is the generation and evaluation of adversarial perturbations.
提供机构：
ILSVRC
