1. [Publications](/index.php/publications) 2. Loss Functions for Image Restoration with Neural Networks # Loss Functions for Image Restoration with Neural Networks ![](/sites/default/files/styles/wide/public/pubs/2017-03_Loss-Functions-for/teaser.png?itok=kjWlOjnf) Neural networks are becoming central in several areas of computer vision and image processing and different architectures have been proposed to solve specific problems. The impact of the loss layer of neural networks, however, has not received much attention in the context of image processing: the default and virtually only choice is L2. In this paper, we bring attention to alternative choices for image restoration. In particular, we show the importance of perceptually-motivated losses when the resulting image is to be evaluated by a human observer. We compare the performance of several losses, and propose a novel, differentiable error function. We show that the quality of the results improves significantly with better loss functions, even when the network architecture is left unchanged. ## Authors Hang Zhao (MIT) Orazio Gallo (NVIDIA) [Iuri Frosio](/index.php/person/iuri-frosio) [Jan Kautz](/index.php/person/jan-kautz) ## Publication Date Wednesday, March 1, 2017 ## Published in [IEEE Transactions on Computational Imaging](http://ieeexplore.ieee.org/document/7797130/) ## Research Area [Computational Photography and Imaging](/index.php/research-area/computational-photography-imaging) [Artificial Intelligence and Machine Learning ](/index.php/research-area/machine-learning-artificial-intelligence) ## Uploaded Files [Paper](https://research.nvidia.com/sites/default/files/pubs/2017-03_Loss-Functions-for/NN_ImgProc.pdf "Open file in new window")8.49 MB [Suppl. Material, Visual Comparison](https://research.nvidia.com/sites/default/files/pubs/2017-03_Loss-Functions-for/comparison_tci.pdf "Open file in new window")45.07 MB [Suppl. Material, Backpropagation](https://research.nvidia.com/sites/default/files/pubs/2017-03_Loss-Functions-for/backpropagation_math.pdf "Open file in new window")122.22 KB [Suppl. Material, Numerical Results](https://research.nvidia.com/sites/default/files/pubs/loss-functions-image-restoration-neural-networks/additionalTables_tci.pdf "Open file in new window")9.1 MB ## Copyright This material is posted here with permission of the IEEE. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to .