arXiv Analytics

Sign in

arXiv:1801.03924 [cs.CV]AbstractReferencesReviewsResources

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric

Richard Zhang, Phillip Isola, Alexei A. Efros, Eli Shechtman, Oliver Wang

Published 2018-01-11Version 1

While it is nearly effortless for humans to quickly assess the perceptual similarity between two images, the underlying processes are thought to be quite complex. Despite this, the most widely used perceptual metrics today, such as PSNR and SSIM, are simple, shallow functions, and fail to account for many nuances of human perception. Recently, the deep learning community has found that features of the VGG network trained on the ImageNet classification task has been remarkably useful as a training loss for image synthesis. But how perceptual are these so-called "perceptual losses"? What elements are critical for their success? To answer these questions, we introduce a new Full Reference Image Quality Assessment (FR-IQA) dataset of perceptual human judgments, orders of magnitude larger than previous datasets. We systematically evaluate deep features across different architectures and tasks and compare them with classic metrics. We find that deep features outperform all previous metrics by huge margins. More surprisingly, this result is not restricted to ImageNet-trained VGG features, but holds across different deep architectures and levels of supervision (supervised, self-supervised, or even unsupervised). Our results suggest that perceptual similarity is an emergent property shared across deep visual representations.

Comments: Code and data available at
Categories: cs.CV, cs.GR
Related articles: Most relevant | Search more
arXiv:1610.07940 [cs.CV] (Published 2016-10-25)
End-to-end Learning of Deep Visual Representations for Image Retrieval
arXiv:1704.05796 [cs.CV] (Published 2017-04-19)
Network Dissection: Quantifying Interpretability of Deep Visual Representations
arXiv:1804.02543 [cs.CV] (Published 2018-04-07)
Not quite unreasonable effectiveness of machine learning algorithms