CINIC-10 is not ImageNet or CIFAR-10

2 Oct 2018Luke N. DarlowElliot J. CrowleyAntreas AntoniouAmos J. Storkey

In this brief technical report we introduce the CINIC-10 dataset as a plug-in extended alternative for CIFAR-10. It was compiled by combining CIFAR-10 with images selected and downsampled from the ImageNet database... (read more)

PDF Abstract

Evaluation results from the paper


Task Dataset Model Metric name Metric value Global rank Compare
Image Classification CINIC-10 VGG-16 Accuracy 87.77 # 4
Image Classification CINIC-10 ResNet-18 Accuracy 90.27 # 3
Image Classification CINIC-10 DenseNet-121 Accuracy 91.26 # 2
Image Classification CINIC-10 ResNeXt29_2x64d Accuracy 91.45 # 1