ImageNet-D
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/chenshuang-zhang/imagenet_d
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为ImageNet-D（背景子集），包含了一些带有误导性背景的图片。它是评估模型性能的标准基准，通常会使用准确率和最差群体准确率这两个指标。该数据集的任务是在误导性背景线索下的物体分类。
The dataset, named ImageNet-D (Background Subset), consists of images with misleading backgrounds. It serves as a standard benchmark for evaluating model performance, with accuracy and worst-group accuracy being the two commonly adopted evaluation metrics. Its core task is object classification under misleading background cues.
