CINIC-10

CINIC-10 是用于图像分类的数据集。它总共有 270,000 张图像，是 CIFAR-10 的 4.5 倍。它由两个不同的来源构成：ImageNet 和 CIFAR-10。具体来说，它被编译为 CIFAR-10 和 ImageNet 之间的桥梁。它被分成三个相等的子集——训练、验证和测试——每个子集包含 90,000 张图像。

CINIC-10 is a dataset for image classification. It contains a total of 270,000 images, 4.5 times that of CIFAR-10. It comprises two distinct sources: ImageNet and CIFAR-10. Specifically, it was developed as a bridge between CIFAR-10 and ImageNet. It is partitioned into three equal subsets: training, validation, and test, each consisting of 90,000 images.

搜集汇总

背景与挑战

背景概述

CINIC-10是一个图像分类数据集，包含27万张图像，规模为CIFAR-10的4.5倍，由ImageNet和CIFAR-10组合构成，旨在作为两者之间的桥梁。该数据集被均匀划分为训练、验证和测试三个子集，每个子集包含9万张图像。