ImageNet-Cartoon and ImageNet-Drawing: two domain shift datasets for ImageNet

<p>Benchmarking the robustness to distribution shifts traditionally relies on dataset collection which is typically laborious and expensive, in particular for datasets with a large number of classes like ImageNet. An exception to this procedure is ImageNet-C (Hendrycks &amp; Dietterich, 2019), a dataset created by applying common real-world corruptions at different levels of intensity to the (clean) ImageNet images. Inspired by this work, we introduce ImageNet-Cartoon and ImageNet-Drawing, two datasets constructed by converting ImageNet images into cartoons and colored pencil drawings, using a GAN framework (Wang &amp; Yu, 2020) and simple image processing (Lu et al., 2012), respectively.</p> <p>This repository contains ImageNet-Cartoon and ImageNet-Drawing.&nbsp;Checkout the&nbsp;<a href="https://github.com/oberman-lab/imagenet-shift">official GitHub Repo</a>&nbsp;for the code on how to reproduce the datasets.</p> <p>If you find this useful in your research, please consider citing:</p> <p>&nbsp; &nbsp; @inproceedings{imagenetshift,<br> &nbsp; &nbsp; &nbsp; title={ImageNet-Cartoon and ImageNet-Drawing: two domain shift datasets for ImageNet},<br> &nbsp; &nbsp; &nbsp; author={Tiago Salvador and Adam M. Oberman},<br> &nbsp; &nbsp; &nbsp; booktitle={ICML Workshop on Shift happens: Crowdsourcing metrics and test datasets beyond ImageNet.},<br> &nbsp; &nbsp; &nbsp; year={2022}<br> &nbsp; &nbsp; }</p>