CINIC-10

Dataset Information
Modalities
Images
Introduced
2018
License
Custom
Homepage
Official Website
Contents

Overview

CINIC-10 is a dataset for image classification. It has a total of 270,000 images, 4.5 times that of CIFAR-10. It is constructed from two different sources: ImageNet and CIFAR-10. Specifically, it was compiled as a bridge between CIFAR-10 and ImageNet. It is split into three equal subsets - train, validation, and test - each of which contain 90,000 images.

Source: Group Knowledge Transfer:Collaborative Training of Large CNNs on the Edge
Image Source: https://arxiv.org/abs/1810.03505

Variants: CINIC-10

Associated Benchmarks

This dataset is used in 3 benchmarks:

Recent Benchmark Submissions

Task Model Paper Date
Image Classification VIT-L/16 (Spinal FC, Background) Reduction of Class Activation Uncertainty … 2023-05-05
Image Classification efficient adaptive ensembling Efficient Adaptive Ensembling for Image … 2022-06-15
Neural Architecture Search NAT-M2 Neural Architecture Transfer 2020-05-12
Neural Architecture Search NAT-M1 Neural Architecture Transfer 2020-05-12
Image Classification NAT-M3 Neural Architecture Transfer 2020-05-12
Image Classification NAT-M2 Neural Architecture Transfer 2020-05-12
Image Classification NAT-M1 Neural Architecture Transfer 2020-05-12
Neural Architecture Search NAT-M4 Neural Architecture Transfer 2020-05-12
Neural Architecture Search NAT-M3 Neural Architecture Transfer 2020-05-12
Image Classification ResNeXt29_2x64d CINIC-10 is not ImageNet or … 2018-10-02
Image Classification DenseNet-121 CINIC-10 is not ImageNet or … 2018-10-02
Image Classification ResNet-18 CINIC-10 is not ImageNet or … 2018-10-02
Image Classification VGG-16 CINIC-10 is not ImageNet or … 2018-10-02

Research Papers

Recent papers with results on this dataset:

External Links: