mini-ImageNet-LT
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/rahulvigneswaran/tailcalibx
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是对mini-ImageNet数据集的扩展，通过特定的不平衡因子进行重采样，形成了长尾分布版本。该数据集包含100个类别，所有图像的尺寸为84x84像素。规模上，训练集包含48,000张图像，验证集和测试集各包含6,000张图像。该数据集的任务是进行图像分类。
This dataset is an extended variant of the mini-ImageNet dataset, resampled using a specific imbalance factor to create a long-tailed distribution version. It consists of 100 categories, with all images having a resolution of 84×84 pixels. In terms of scale, the training set contains 48,000 images, while the validation and test sets each include 6,000 images. The task for this dataset is image classification.
提供机构：
mini-ImageNet
